Discussion:
GSOC 2012
(too old to reply)
William Stein
2012-02-08 03:48:39 UTC
Permalink
Hi,

Does anybody want to help put together a GSoC application for Sage for 2012?

http://code.google.com/soc/

The application deadline is March 9.

So far, I think we've applied 5 times to have Sage as a mentoring
organizing, and been denied every time. I think there is no feedback
about why we are denied (maybe they think we already have too much NSF
funding?). Also, other similar projects such as R, Sympy, PlanetMath,
etc., have often been accepted as mentoring organizations. However,
I don't think being denied every year is a reason to stop trying,
because (1) our project is better than many of the projects Google
chooses (they are just making a mistake by not choosing us), and (2)
even if they don't choose us, we can propose our project ideas to
other mentoring organizations. Regarding (2) though, it can be
frustrating -- e.g., I felt we had an excellent proposal for a
mentoring organization one of the years Sage was denied, and the
organization decided against funding it because the developers didn't
know us personally; their "no" was not based on weaknesses of the
project itself, which I found frustrating. So it's best if we are a
mentoring organization.

-- William
--
William Stein
Professor of Mathematics
University of Washington
http://wstein.org
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Jason Grout
2012-02-08 03:55:42 UTC
Permalink
Post by William Stein
Hi,
Does anybody want to help put together a GSoC application for Sage for 2012?
http://code.google.com/soc/
The application deadline is March 9.
So far, I think we've applied 5 times to have Sage as a mentoring
organizing, and been denied every time. I think there is no feedback
about why we are denied (maybe they think we already have too much NSF
funding?). Also, other similar projects such as R, Sympy, PlanetMath,
etc., have often been accepted as mentoring organizations. However,
I don't think being denied every year is a reason to stop trying,
because (1) our project is better than many of the projects Google
chooses (they are just making a mistake by not choosing us), and (2)
even if they don't choose us, we can propose our project ideas to
other mentoring organizations. Regarding (2) though, it can be
frustrating -- e.g., I felt we had an excellent proposal for a
mentoring organization one of the years Sage was denied, and the
organization decided against funding it because the developers didn't
know us personally; their "no" was not based on weaknesses of the
project itself, which I found frustrating. So it's best if we are a
mentoring organization.
So are you saying we shouldn't, for example, submit under the Python
Software Foundation umbrella?

I'd be willing to mentor, for example, a notebook project, since that's
what I'll be working on most of the summer. I'll already have several
students hopefully working with me on the notebook, or graphics (webgl,
here we come :), etc.

Jason
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
William Stein
2012-02-08 04:00:30 UTC
Permalink
Post by Jason Grout
Post by William Stein
Hi,
Does anybody want to help put together a GSoC application for Sage for 2012?
   http://code.google.com/soc/
The application deadline is March 9.
So far, I think we've applied 5 times to have Sage as a mentoring
organizing, and been denied every time.  I think there is no feedback
about why we are denied (maybe they think we already have too much NSF
funding?).  Also, other similar projects such as R, Sympy, PlanetMath,
etc., have often been accepted as mentoring organizations.    However,
I don't think being denied every year is a reason to stop trying,
because (1) our project is better than many of the projects Google
chooses (they are just making a mistake by not choosing us), and (2)
even if they don't choose us, we can propose our project ideas to
other mentoring organizations.   Regarding (2) though, it can be
frustrating -- e.g., I felt we had an excellent proposal for a
mentoring organization one of the years Sage was denied, and the
organization decided against funding it because the developers didn't
know us personally; their "no" was not based on weaknesses of the
project itself, which I found frustrating.   So it's best if we are a
mentoring organization.
So are you saying we shouldn't, for example, submit under the Python
Software Foundation umbrella?
If we come up with good projects that would make sense for PSF *and*
are denied as a mentoring organization, then we should summit there.
In my remark above, the mentoring org was not PSF.
Post by Jason Grout
I'd be willing to mentor, for example, a notebook project, since that's what
I'll be working on most of the summer.  I'll already have several students
hopefully working with me on the notebook, or graphics (webgl, here we come
:), etc.
Excellent. I am also willing to mentor a project on the notebook.

Or something on implementing mathematical algorithms.
Post by Jason Grout
Jason
--
To unsubscribe from this group, send an email to
For more options, visit this group at
http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
--
William Stein
Professor of Mathematics
University of Washington
http://wstein.org
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
kcrisman
2012-02-08 04:20:31 UTC
Permalink
Post by Jason Grout
I'd be willing to mentor, for example, a notebook project, since that's what
I'll be working on most of the summer.  I'll already have several students
hopefully working with me on the notebook, or graphics (webgl, here we come
:), etc.
Excellent.  I am also willing to mentor a project on the notebook.
Or something on implementing mathematical algorithms.
Looking at Sympy's GSOC 2011 stuff, there is a lot of nontrivial math
in it - so perhaps GSOC is getting more interested in funding more
mathematical stuff (which was bandied about as a possible reason for
previous denials).

In fact, Sympy is looking like it's getting pretty impressive.

I hesitate to say I would be a good mentor, but there are a lot of
things in symbolics and graphics that would be appropriate for this
that I'd like to try with some of my students. Especially piecewise
functions and such. Continuing nontrivial Geogebra integration could
be another very appropriate one.

- kcrisman
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
mmarco
2012-02-08 10:15:23 UTC
Permalink
Post by kcrisman
Continuing nontrivial Geogebra integration could
be another very appropriate one.
- kcrisman
Big +1 to that.
Post by kcrisman
* webwork/sage integration
I would also add integration with moodle (maybe that would include
ldap?)


There is also some mathematical stuff that i would like to see in
Sage, but i am not sure if it would be appropiate for a GSOC project.
Just to name a few that i would be interested in:
Infinite groups (free, finitelly presented, braid groups...)
Algebraic geometry (improve the schemes implementation, blowups,
sheaves, characteristic classes...)
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Michael Orlitzky
2012-02-08 13:12:28 UTC
Permalink
Post by kcrisman
I hesitate to say I would be a good mentor, but there are a lot of
things in symbolics and graphics that would be appropriate for this
that I'd like to try with some of my students. Especially piecewise
functions and such. Continuing nontrivial Geogebra integration could
be another very appropriate one.
If you're serious about fixing piecewise,

* I'll be a student until at least May (this is fine by the GSOC
rules).

* I'll have the summer off.

* I've been working on splines for the past few years, and am
familiar with all of the reasons why piecewise needs fixing.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Christopher Swenson
2012-02-08 19:43:15 UTC
Permalink
I thought I would not that, to whomever is going to write the application,
be sure to list me down as a Google supporter, since I think it would be
great to have a Sage GSoC project. There should be a section along the
lines of "vouchers from Google and other large organizations".

If I had a bit more experience with Sage development, I would have been
happy to mentor someone as well. :)

--Christopher
Post by Michael Orlitzky
Post by kcrisman
I hesitate to say I would be a good mentor, but there are a lot of
things in symbolics and graphics that would be appropriate for this
that I'd like to try with some of my students. Especially piecewise
functions and such. Continuing nontrivial Geogebra integration could
be another very appropriate one.
If you're serious about fixing piecewise,
* I'll be a student until at least May (this is fine by the GSOC
rules).
* I'll have the summer off.
* I've been working on splines for the past few years, and am
familiar with all of the reasons why piecewise needs fixing.
--
For more options, visit this group at http://groups.google.com/**
group/sage-devel <http://groups.google.com/group/sage-devel>
URL: http://www.sagemath.org
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
William Stein
2012-02-08 19:49:13 UTC
Permalink
On Wed, Feb 8, 2012 at 11:43 AM, Christopher Swenson
Post by Christopher Swenson
I thought I would not that, to whomever is going to write the application,
be sure to list me down as a Google supporter, since I think it would be
great to have a Sage GSoC project.  There should be a section along the
lines of "vouchers from Google and other large organizations".
In case it isn't obvious, Chris works at Google.

William
Post by Christopher Swenson
If I had a bit more experience with Sage development, I would have been
happy to mentor someone as well. :)
--Christopher
Post by Michael Orlitzky
Post by kcrisman
I hesitate to say I would be a good mentor, but there are a lot of
things in symbolics and graphics that would be appropriate for this
that I'd like to try with some of my students.  Especially piecewise
functions and such.  Continuing nontrivial Geogebra integration could
be another very appropriate one.
If you're serious about fixing piecewise,
 * I'll be a student until at least May (this is fine by the GSOC
   rules).
 * I'll have the summer off.
 * I've been working on splines for the past few years, and am
   familiar with all of the reasons why piecewise needs fixing.
--
To unsubscribe from this group, send an email to
For more options, visit this group at
http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
--
To unsubscribe from this group, send an email to
For more options, visit this group at
http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
--
William Stein
Professor of Mathematics
University of Washington
http://wstein.org
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Harald Schilly
2012-02-08 19:50:36 UTC
Permalink
There should be a section along the lines of "vouchers from Google and other
large organizations".
yes, there is in deed such a section. thanks for your support :)

H
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Michael Orlitzky
2012-02-12 04:39:32 UTC
Permalink
I've started this:

http://wiki.sagemath.org/PiecewiseSymbolicSEP

It's basically a brain dump at this point, but I can go back and clean
up specific ideas now with less overhead.

I've also added a link and a few paragraphs to the GSoC proposal.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
John H Palmieri
2012-02-12 06:33:36 UTC
Permalink
Post by Michael Orlitzky
http://wiki.sagemath.org/PiecewiseSymbolicSEP
It's basically a brain dump at this point, but I can go back and clean
up specific ideas now with less overhead.
I've also added a link and a few paragraphs to the GSoC proposal.
Would it also be possible and worthwhile to implement periodic functions at
the same time? That is, you could define a function on some interval [a,b]
(or [a,b) or ...) and declare it to be periodic with period b-a, for
instance.
--
John
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Robert Bradshaw
2012-02-12 07:11:32 UTC
Permalink
On Sat, Feb 11, 2012 at 10:33 PM, John H Palmieri
Post by John H Palmieri
   http://wiki.sagemath.org/PiecewiseSymbolicSEP
It's basically a brain dump at this point, but I can go back and clean
up specific ideas now with less overhead.
I've also added a link and a few paragraphs to the GSoC proposal.
Would it also be possible and worthwhile to implement periodic functions at
the same time?  That is, you could define a function on some interval [a,b]
(or [a,b) or ...) and declare it to be periodic with period b-a, for
instance.
I think such functions should be implemented based on predicates
rather than intervals. Then they could support this as well as being
multi-variate, etc. Unfortunately the mod operator doesn't seem to
work with symbolics (could be adde to the SEP)

sage: x % 1
...
TypeError: unable to convert x (=x) to an integer

And we also have the ugly

sage: 0 < x < 5
0 < x

and

sage: var('y,x')
(y, x)
sage: (0 < x) & (0 < y)
...
TypeError: unsupported operand type(s) for &:
'sage.symbolic.expression.Expression' and
'sage.symbolic.expression.Expression'

Part of this SEP could be to make both of these work (if possible, the
first might be impossible due to shortcutting and bool(expr) returning
False by default). Note that if the predicates are evaluated in a
cascading manner, this wouldn't be needed as often.

Another complication is detecting boundaries (e.g. for
differentiability tests) for arbitrary predicates, but this could be
done for inequalities.

- Robert
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Michael Orlitzky
2012-02-12 15:12:53 UTC
Permalink
Post by Michael Orlitzky
http://wiki.sagemath.org/PiecewiseSymbolicSEP
<http://wiki.sagemath.org/PiecewiseSymbolicSEP>
It's basically a brain dump at this point, but I can go back and clean
up specific ideas now with less overhead.
I've also added a link and a few paragraphs to the GSoC proposal.
Would it also be possible and worthwhile to implement periodic functions
at the same time? That is, you could define a function on some interval
[a,b] (or [a,b) or ...) and declare it to be periodic with period b-a,
for instance.
Assuming we get piecewise right, it *sounds* easy to do. I've added it
to the SEP as a feature that could be built on top of piecewise.

One goal would be to define the piecewise functions based on predicates,
so we could have e.g. a periodic(f, interval, period) constructor that
converts your interval/period to a predicate and returns the
corresponding piecewise function.

I can see it getting hairy if we try to generalize periodic functions,
too, but period/interval are easy enough to work with.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
rjf
2012-02-13 17:08:40 UTC
Permalink
Post by Michael Orlitzky
I can see it getting hairy if we try to generalize periodic functions,
too, but period/interval are easy enough to work with.
There are a whole bunch of challenging problems, though in many
cases a half-backed trivial solution to the easy parts can be
hacked together.
You can see how Maxima packages do stuff related to piecewise
definitions, periodic and oscillatory functions (esp. integrating)
and see what Mathematica does. I wrote a paper on symmetric/ anti-
symmetric
etc functions and simplification.

RJF
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Volker Braun
2012-02-08 04:11:32 UTC
Permalink
Maybe we can tie the Android app into the notebook work? I could help
mentor a student working on Android stuff, maybe that would help our GSOC
application.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
William Stein
2012-02-08 04:15:19 UTC
Permalink
Maybe we can tie the Android app into the notebook work? I could help mentor
a student working on Android stuff, maybe that would help our GSOC
application.
That's an *extremely* good idea!

Maybe we should have "The Sage Notebook" as the mentoring organization
instead of Sage, given that every project idea so far has involved it?

William
--
To unsubscribe from this group, send an email to
For more options, visit this group at
http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
--
William Stein
Professor of Mathematics
University of Washington
http://wstein.org
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Jason Grout
2012-02-08 04:27:09 UTC
Permalink
Post by William Stein
Maybe we can tie the Android app into the notebook work? I could help mentor
a student working on Android stuff, maybe that would help our GSOC
application.
That's an *extremely* good idea!
Maybe we should have "The Sage Notebook" as the mentoring organization
instead of Sage, given that every project idea so far has involved it?
Here are some more ideas:

* get interacts working in the IPython html notebook (not exactly Sage,
but very closely related, and likely far more reaching than a Sage
notebook project)

* python->javascript translator so we can have *real* interaction (e.g.,
convert a symbolic expression to javascript, so the interact is totally
computed in the browser, for simple interacts)

* the usual: folders/tags/some method for organization, migrate to
vastly different architecture for notebook (like ipython 0.12 or the
sage cell architecture), webgl frontend for 3d graphics with all the
trappings

Sage projects:

* get fast_callable into shape, including a fortran backend (I was
really impressed with Oscar's demonstrations the other day). See
http://trac.sagemath.org/sage_trac/ticket/5572; the sky is the limit for
this as far as low-level expression optimizations too)

* overhaul 2d graphics to be consistent, take advantage of matplotlib
much more, etc. Introduce svg or html5 frontends for matplotlib that
make interactive browser graphics easier (like interacts)

* webwork/sage integration

* coercion system printing framework, like David Roe has long advocated

Thanks,

Jason
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
kcrisman
2012-02-08 05:17:40 UTC
Permalink
Post by Jason Grout
* overhaul 2d graphics to be consistent, take advantage of matplotlib
much more, etc.  Introduce svg or html5 frontends for matplotlib that
make interactive browser graphics easier (like interacts)
These would be really cool.
Post by Jason Grout
* webwork/sage integration
There are real people already working on this, and WW is definitely a
good partner for this - great idea. Cc:ing a few relevant people in
case WW would be interested in cosponsoring that particular idea.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Jason Aubrey
2012-02-08 05:26:42 UTC
Permalink
Post by kcrisman
Post by Jason Grout
* overhaul 2d graphics to be consistent, take advantage of matplotlib
much more, etc. Introduce svg or html5 frontends for matplotlib that
make interactive browser graphics easier (like interacts)
These would be really cool.
Post by Jason Grout
* webwork/sage integration
There are real people already working on this, and WW is definitely a
good partner for this - great idea. Cc:ing a few relevant people in
case WW would be interested in cosponsoring that particular idea.
Yes, we are certainly interested! <I boldly assert on behalf of Mike and
John ;) >

JasonA
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
John Travis
2012-02-08 05:28:42 UTC
Permalink
+1

I had to go look up what GSOC was first ...

Jt
Post by Jason Aubrey
Post by kcrisman
Post by Jason Grout
* overhaul 2d graphics to be consistent, take advantage of matplotlib
much more, etc. Introduce svg or html5 frontends for matplotlib that
make interactive browser graphics easier (like interacts)
These would be really cool.
Post by Jason Grout
* webwork/sage integration
There are real people already working on this, and WW is definitely a
good partner for this - great idea. Cc:ing a few relevant people in
case WW would be interested in cosponsoring that particular idea.
Yes, we are certainly interested! <I boldly assert on behalf of Mike and
John ;) >
JasonA
______________________________________
CONFIDENTIALITY NOTE: This communication may contain confidential information.
If you are not the intended recipient or if you are not authorized to receive it, please
notify and return the message to the sender. Unauthorized reviewing, forwarding,
copying, distributing or using this information is strictly prohibited.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Martin Albrecht
2012-02-08 10:00:50 UTC
Permalink
Hi,

I could be tempted to mentor as well, but the group of potential beneficiaries
of the stuff I have in mind is much smaller than notebook stuff or the Android
app. Anyway, here it goes:

a) Dense linear algebra over extension fields. I'm meeting Burcin tomorrow to
work on this, so we'll have to decide whether it makes sense as a GSOC project
or not. But the idea is to use tuples of matrices to represent matrices over
extension fields ... and to have that reasonably generic in Sage so that it
works for GF(p^k) and number fields, say.

b) Write a decent interface to CryptMiniSat or any other SAT solver for Sage.
By decent I mean something on Cython level, allowing direct access to the
solver, e.g. getting learned clauses out etc. Also, bonus brownie points for
ANF 2 CNF conversion (there's decent stuff out there one could wrap).

c) M1RI, that's the code Tom wrote for dense linear algebra over GF(3) -
GF(7). As far as I understand it, it's a bunch of Sage worksheets at the
moment, i.e. a proof of concept. Turning this into a proper library would be
awesome. It requires C skills but most of the linear algebra business has been
solved before, i.e., I could give decent advise on how implement this or that
function etc.
Post by William Stein
Hi,
Does anybody want to help put together a GSoC application for Sage for 2012?
http://code.google.com/soc/
The application deadline is March 9.
So far, I think we've applied 5 times to have Sage as a mentoring
organizing, and been denied every time. I think there is no feedback
about why we are denied (maybe they think we already have too much NSF
funding?). Also, other similar projects such as R, Sympy, PlanetMath,
etc., have often been accepted as mentoring organizations. However,
I don't think being denied every year is a reason to stop trying,
because (1) our project is better than many of the projects Google
chooses (they are just making a mistake by not choosing us), and (2)
even if they don't choose us, we can propose our project ideas to
other mentoring organizations. Regarding (2) though, it can be
frustrating -- e.g., I felt we had an excellent proposal for a
mentoring organization one of the years Sage was denied, and the
organization decided against funding it because the developers didn't
know us personally; their "no" was not based on weaknesses of the
project itself, which I found frustrating. So it's best if we are a
mentoring organization.
-- William
Cheers,
Martin

--
name: Martin Albrecht
_pgp: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x8EF0DC99
_otr: 47F43D1A 5D68C36F 468BAEBA 640E8856 D7951CCF
_www: http://martinralbrecht.wordpress.com/
_jab: ***@jabber.ccc.de
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Minh Nguyen
2012-02-08 10:36:52 UTC
Permalink
Hi Martin,

On Wed, Feb 8, 2012 at 9:00 PM, Martin Albrecht
Post by Martin Albrecht
c) M1RI, that's the code Tom wrote for dense linear algebra over GF(3) -
GF(7). As far as I understand it, it's a bunch of Sage worksheets at the
moment, i.e. a proof of concept.
Where are these Sage worksheets?
--
Regards,
Minh Van Nguyen
http://sage.math.washington.edu/home/mvngu/
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Harald Schilly
2012-02-08 11:04:55 UTC
Permalink
I've started to write a GSOC 2012 application on the day we got rejected in
2011. Whoever wants to help, i'll add you to the google docs document. I'm
fine with any contributions or even submitting this as the "notebook" and
not sage itself. I also got some feedback from the last submissions
directly from the people who were evaluating it (in a irc session). So, I
can give you additional feedback on what to change and what do not change.
I think, this year our chances are higher than last year, because last year
they focused on smaller projects and it might be logically to focus on
bigger ones this time.

H
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Dr. David Kirkby
2012-02-09 00:12:17 UTC
Permalink
Post by William Stein
Hi,
Does anybody want to help put together a GSoC application for Sage for 2012?
http://code.google.com/soc/
The application deadline is March 9.
So far, I think we've applied 5 times to have Sage as a mentoring
organizing, and been denied every time. I think there is no feedback
about why we are denied (maybe they think we already have too much NSF
funding?).
Lack of feedback is really annoying. Can't you get any unofficial feedback via
telephone calls rather than email?
Post by William Stein
Also, other similar projects such as R, Sympy, PlanetMath,
etc., have often been accepted as mentoring organizations. However,
I don't think being denied every year is a reason to stop trying,
because (1) our project is better than many of the projects Google
chooses (they are just making a mistake by not choosing us),
Perhaps that's not the wisest thing to state in public.

I've had a couple of theories about why applications might fail.

1) Too mathematical, though that theory seems to have been dismissed, as I
gather other heavy maths has been funded.

2) The Sage development process is not exactly a shining example of best
practice in software engineering. I once suggested you purchased some books on
software engineering and handed out free copies to some of your developers,
since its clear some don't have a clue.

It's remotely possible Google see this, and would rather someone mentored in a
different environment, which at least appears a bit different, even if in fact
it is no less chaotic.

Dave
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
William Stein
2012-02-09 00:19:25 UTC
Permalink
On Wed, Feb 8, 2012 at 4:12 PM, Dr. David Kirkby
Post by Dr. David Kirkby
Post by William Stein
Hi,
Does anybody want to help put together a GSoC application for Sage for 2012?
   http://code.google.com/soc/
The application deadline is March 9.
So far, I think we've applied 5 times to have Sage as a mentoring
organizing, and been denied every time.  I think there is no feedback
about why we are denied (maybe they think we already have too much NSF
funding?).
Lack of feedback is really annoying. Can't you get any unofficial feedback
via telephone calls rather than email?
My post on this list resulted in unofficial off-list feedback.
Post by Dr. David Kirkby
Post by William Stein
Also, other similar projects such as R, Sympy, PlanetMath,
etc., have often been accepted as mentoring organizations.    However,
I don't think being denied every year is a reason to stop trying,
because (1) our project is better than many of the projects Google
chooses (they are just making a mistake by not choosing us),
Perhaps that's not the wisest thing to state in public.
Dang, I guess the secret is out now.

-- William
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
daly
2012-02-09 00:31:06 UTC
Permalink
Post by Dr. David Kirkby
Post by William Stein
Hi,
Does anybody want to help put together a GSoC application for Sage for 2012?
http://code.google.com/soc/
The application deadline is March 9.
So far, I think we've applied 5 times to have Sage as a mentoring
organizing, and been denied every time. I think there is no feedback
about why we are denied (maybe they think we already have too much NSF
funding?).
Lack of feedback is really annoying. Can't you get any unofficial feedback via
telephone calls rather than email?
Axiom was chosen as one of the project for the first GSoC
for about 24 hours, at which point it was dropped. I never
did find out why.

We did have projects under the LispNYC umbrella organization
due to Goodman's efforts.
Post by Dr. David Kirkby
Post by William Stein
Also, other similar projects such as R, Sympy, PlanetMath,
etc., have often been accepted as mentoring organizations. However,
I don't think being denied every year is a reason to stop trying,
because (1) our project is better than many of the projects Google
chooses (they are just making a mistake by not choosing us),
Perhaps that's not the wisest thing to state in public.
I've had a couple of theories about why applications might fail.
1) Too mathematical, though that theory seems to have been dismissed, as I
gather other heavy maths has been funded.
Axiom got selected but the effort was for a web-based notebook-style
interface which is clearly not mathematical and, as you already know,
is so wildly futuristic that it will likely never be achieved :-)
Post by Dr. David Kirkby
2) The Sage development process is not exactly a shining example of best
practice in software engineering. I once suggested you purchased some books on
software engineering and handed out free copies to some of your developers,
since its clear some don't have a clue.
So stretch a bit and propose a literate programming Sage example.
I'd be willing to act as a Sage mentor for that.
Post by Dr. David Kirkby
It's remotely possible Google see this, and would rather someone mentored in a
different environment, which at least appears a bit different, even if in fact
it is no less chaotic.
I suspect that, like all things Google, they just get swamped by
applications. Google funding applications are to NSF applications
like Chatroulette is to dating. :-) It wouldn't surprise me to find
out that they have automated the selection criteria just to deal
with the volume.

Tim Daly
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Julien Puydt
2012-02-09 12:42:10 UTC
Permalink
Post by Dr. David Kirkby
I've had a couple of theories about why applications might fail.
1) Too mathematical, though that theory seems to have been dismissed,
as I gather other heavy maths has been funded.
I don't think so.
Post by Dr. David Kirkby
2) The Sage development process is not exactly a shining example of
best practice in software engineering. I once suggested you purchased
some books on software engineering and handed out free copies to some
of your developers, since its clear some don't have a clue.
It's remotely possible Google see this, and would rather someone
mentored in a different environment, which at least appears a bit
different, even if in fact it is no less chaotic.
That second point convinces me much more ; consider :

1. Is it available readily in most distributions? No.

2. Does ./configure && make && make install give something in a
reasonable time? Uh. No configure script?! Ah, make works... what?! It
is compiling bzip2!? Stop!

Snark on #sagemath
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Harald Schilly
2012-02-09 12:57:45 UTC
Permalink
Post by Julien Puydt
Post by Dr. David Kirkby
I've had a couple of theories about why applications might fail.
2) The Sage development process is not exactly a shining example of
best practice in software engineering.
1. Is it available readily in most distributions? No.
2. Does ./configure && make && make install give something in a
reasonable time? Uh. ...
Well, from what I read so far in the last years, this is not a
criteria at all. I think you are simply trapped in the human thinking
process, that you try to find a reason where no reason is. There is a
lot of luck involved and if sage doesn't fit on a very rough level, it
won't be selected.
It's like if you play a random game, and you think it "must happen
now" and if it still does not, you blame it on wearing the wrong
socks…

H
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
William Stein
2012-02-09 15:27:51 UTC
Permalink
Post by Harald Schilly
Post by Julien Puydt
Post by Dr. David Kirkby
I've had a couple of theories about why applications might fail.
2) The Sage development process is not exactly a shining example of
best practice in software engineering.
1. Is it available readily in most distributions? No.
2. Does ./configure && make && make install give something in a
reasonable time? Uh. ...
Well, from what I read so far in the last years, this is not a
criteria at all. I think you are simply trapped in the human thinking
process, that you try to find a reason where no reason is. There is a
lot of luck involved and if sage doesn't fit on a very rough level, it
won't be selected.
It's like if you play a random game, and you think it "must happen
now" and if it still does not, you blame it on wearing the wrong
socks…
+1
Post by Harald Schilly
H
--
To unsubscribe from this group, send an email to
For more options, visit this group at
http://groups.google.com/group/sage-devel
Post by Harald Schilly
URL: http://www.sagemath.org
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Rob Beezer
2012-02-10 05:51:24 UTC
Permalink
Post by Dr. David Kirkby
since its clear some don't have a clue.
"Perhaps that's not the wisest thing to state in public."

Rob
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Keshav Kini
2012-02-10 06:54:37 UTC
Permalink
Well, to be fair, it's generally less wise to state in public how much
others suck than to state in public how much we ourselves suck :)

-Keshav

----
Join us in #sagemath on irc.freenode.net !
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
rjf
2012-02-12 15:06:48 UTC
Permalink
Microsoft and Google/Motorola are suing each other.
Do you get money from Microsoft?
Just a thought.

You could propose to Google to port Sage to run on Windows/ natively,
not.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Dr. David Kirkby
2012-02-13 08:51:29 UTC
Permalink
Post by rjf
Microsoft and Google/Motorola are suing each other.
Do you get money from Microsoft?
Just a thought.
You could propose to Google to port Sage to run on Windows/ natively,
not.
I believe a complete native port would be an almost impossibility. I don't think
many developers are keen on Windoze, so it would be less easy to attract
developers unless they were paid to do it.

I must admit, I would have thought a Cygwin port easier than it seems to be. I
recall at one time wondering whether the Solaris or Cygwin port would be
completed first, but Solaris got there several years ago, and the Cgywin port
seems less likely to be completed now than it did a few years ago.

Microsoft were at some time funding a Windoze port of Sage, but I believe the
amount of funding they gave was far too small to make the slightest dent in the
problem.

Dave
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
rjf
2012-02-13 17:22:22 UTC
Permalink
Post by Dr. David Kirkby
  Microsoft and Google/Motorola are suing each other.
Do you get money from Microsoft?
Just a thought.
You could propose to Google to port Sage to run on Windows/ natively,
not.
I believe a complete native port would be an almost impossibility. I don't think
many developers are keen on Windoze, so it would be less easy to attract
developers unless they were paid to do it.
So ask for money to pay them not to do it, from Google.

Frankly, I think a person competent in Windows technology could do
this.
In my own experience, Maxima runs fine under windows.
Also in my experience, GMP and MPFR run under Windows, but maybe
require additional non-free software development environments.
This is not a problem if you drop the requirement that every recipient
of Sage must be able to COMPILE stuff locally. Just have one person
compile the stuff once and distribute dll files.

I suppose
it is possible that some Sage propeller-heads have written python or C
code that cannot be compiled under windows, but it seems to me more
likely that it is a question of finding a skilled person, and not so
much of problematical code. A skilled person who cares to look at
code base, and has a spare week or two.
Post by Dr. David Kirkby
I must admit, I would have thought a Cygwin port easier than it seems to be. I
recall at one time wondering whether the Solaris or Cygwin port would be
completed first, but Solaris got there several years ago, and the Cgywin port
seems less likely to be completed now than it did a few years ago.
Microsoft were at some time funding a Windoze port of Sage, but I believe the
amount of funding they gave was far too small to make the slightest dent in the
problem.
Money misspent on unskilled labor, perhaps.

Incidentally, I am by no means suggesting that I have the needed
skills.
(I think I can identify someone who could do it, however.)

RJF
Post by Dr. David Kirkby
Dave
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Dima Pasechnik
2012-02-14 02:10:09 UTC
Permalink
Post by rjf
Post by Dr. David Kirkby
  Microsoft and Google/Motorola are suing each other.
Do you get money from Microsoft?
Just a thought.
You could propose to Google to port Sage to run on Windows/ natively,
not.
I believe a complete native port would be an almost impossibility. I don't think
many developers are keen on Windoze, so it would be less easy to attract
developers unless they were paid to do it.
So ask for money to pay them not to do it, from Google.
Frankly, I think a person competent in Windows technology could do
this.
In my own experience, Maxima runs fine under windows.
Also in my experience, GMP and MPFR run under Windows, but maybe
require additional non-free software development environments.
GMP and MPFR is only a small part of the story. There are huge packages
in Sage (not developed by Sage people) that do not run natively under
Windows and are quite non-trivial to port (as they might have their own
tricky GC, use fork and its POSIX-only friends, original developers of
these parts left long ago, etc). E.g. GAP alone is a nontrivial task to
port.
Post by rjf
This is not a problem if you drop the requirement that every recipient
of Sage must be able to COMPILE stuff locally. Just have one person
compile the stuff once and distribute dll files.
this means distributing a seriously crippled system.
Post by rjf
I suppose
it is possible that some Sage propeller-heads have written python or C
code that cannot be compiled under windows, but it seems to me more
likely that it is a question of finding a skilled person, and not so
much of problematical code. A skilled person who cares to look at
code base, and has a spare week or two.
Post by Dr. David Kirkby
I must admit, I would have thought a Cygwin port easier than it seems to be. I
recall at one time wondering whether the Solaris or Cygwin port would be
completed first, but Solaris got there several years ago, and the Cgywin port
seems less likely to be completed now than it did a few years ago.
Microsoft were at some time funding a Windoze port of Sage, but I believe the
amount of funding they gave was far too small to make the slightest dent in the
problem.
Money misspent on unskilled labor, perhaps.
Incidentally, I am by no means suggesting that I have the needed
skills.
(I think I can identify someone who could do it, however.)
RJF
Post by Dr. David Kirkby
Dave
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
rjf
2012-02-14 05:34:17 UTC
Permalink
On Feb 13, 6:10 pm, Dima Pasechnik <***@gmail.com> wrote:
.....
Post by Dima Pasechnik
GMP and MPFR is only a small part of the story. There are huge packages
in Sage (not developed by Sage people) that do not run natively under
Windows and are quite non-trivial to port (as they might have their own
tricky GC, use fork and its POSIX-only friends, original developers of
these parts left long ago, etc). E.g. GAP alone is a nontrivial task to
port.
(RJF) This is not a problem if you drop the requirement that every recipient
of Sage must be able to COMPILE stuff locally.  Just have one person
compile the stuff once and distribute dll files.
this means distributing a seriously crippled system.
How so?
Does that mean that only those people who compile Sage themselves
have a non-crippled system?


If Sage/linux has adopted large subsystems that do not run in windows,
then
those systems can be dropped from the Sage/windows distribution.
Report the
problem as a bug to the author(s) of such systems. When they are fixed
so they can be compiled and run in windows, put them into the windows
distribution. Problem solved.

If you are saying that your home-grown core python Sage stuff cannot
be run in
Windows, then shame on you for turning your back on what is still the
most wide-spread operating system (or family of systems) on home
and office computers.

RJF
Post by Dima Pasechnik
 I suppose
it is possible that some Sage propeller-heads have written python or C
code that cannot be compiled under windows, but it seems to me more
likely that it is a question of finding a skilled person, and not so
much of problematical code.  A skilled person who cares to look at
code base, and has a spare week or two.
Post by Dr. David Kirkby
I must admit, I would have thought a Cygwin port easier than it seems to be. I
recall at one time wondering whether the Solaris or Cygwin port would be
completed first, but Solaris got there several years ago, and the Cgywin port
seems less likely to be completed now than it did a few years ago.
Microsoft were at some time funding a Windoze port of Sage, but I believe the
amount of funding they gave was far too small to make the slightest dent in the
problem.
Money misspent on unskilled labor, perhaps.
Incidentally, I am by no means suggesting that I have the needed
skills.
(I think I can identify someone who could do it, however.)
RJF
Post by Dr. David Kirkby
Dave
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Volker Braun
2012-02-14 06:05:04 UTC
Permalink
Post by rjf
shame on you for turning your back on what is still the
most wide-spread operating system (or family of systems) on home
and office computers.
The most common home computer is a game console. While techincally an XBox
is a windows PC, very few normal Windows programs run on it.

Also, Windows has zero market share in HPC. But if you think that our
primary goal should be to have a native port for gamers and secretaries
then I guess I have no choice but to drop everything else and work on the
Windows API du jour.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Harald Schilly
2012-02-14 10:37:10 UTC
Permalink
Post by Volker Braun
The most common home computer is a game console.
in 2012, i think it's the posix-world of smartphones and tablets,
namely android and iOS. porting to or supporting a growing ecosystem
(i.e. tablets) has more long term benefits than petting a dying
dinosaur.

h
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
David Kirkby
2012-02-14 11:16:04 UTC
Permalink
Post by Volker Braun
The most common home computer is a game console. While techincally an
XBox is a windows PC, very few normal Windows programs run on it.

Some of my code has been run on a Sony Playstation 3, as well as a Cray
supercomputer.

http://atlc.sourceforge.net/
Post by Volker Braun
Also, Windows has zero market share in HPC.
I disagree with that statement.

As you know, I'm not fan of Windoze, but I don't accept that Windoze has no
market share in HPC. There are many HPC applications which support
primarily Windows.

HFSS

http://www.ansoft.com/products/hf/hfss/

is one such example. I was recently speaking to one of their engineers, who
said some customers had simultions taking weeks, so they would get Ansys
(the vendor) to simulate them on more powerful hardware, reducing the time
from weeks to days. HFSS supports both multi-processors on the one machine,
as well as shared processors across a network and shared memory across a
network.

Speaking to one of the salemen, he said Windows was the most common
platform for HFSS. One of the packages that can be used with HFSS (the free
Antenna design kit), is only available on Windows, though the antenna
design kit is certainly not particuly CPU intensive and I doubt a design
would take more than 30 minutes.

Personally I'd consider simulation software which takes days/weeks to
simulate complex engineering problems to fall under the gategory of HPC,
though I don't think there's any formal definition of HPC, so I could not
really argue with you if you said such software was not.

All similar packages to HFSS (EMPro, XCcell, CST etc) support Windows and
Linux. Sometimes Solaris is supported, sometimes OS X too.


Dave
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Volker Braun
2012-02-14 23:01:59 UTC
Permalink
Post by David Kirkby
is one such example. I was recently speaking to one of their engineers,
who said some customers had simultions taking weeks, so they would get
Ansys (the vendor) to simulate them on more powerful hardware, reducing the
time from weeks to days.
If its true that Anysys has a reasonably-sized cluster (say, 100+ nodes)
that they rent out to customers then I'm pretty sure it doesn't run on
Windows, especially if their own code is cross-platform. This is what I
meant. The fact that their code also runs on desktops doesn't make the
desktop a HPC platform.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
François Bissey
2012-02-14 23:24:24 UTC
Permalink
Post by Volker Braun
Post by David Kirkby
is one such example. I was recently speaking to one of their engineers,
who said some customers had simultions taking weeks, so they would get
Ansys (the vendor) to simulate them on more powerful hardware, reducing
the time from weeks to days.
If its true that Anysys has a reasonably-sized cluster (say, 100+ nodes)
that they rent out to customers then I'm pretty sure it doesn't run on
Windows, especially if their own code is cross-platform. This is what I
meant. The fact that their code also runs on desktops doesn't make the
desktop a HPC platform.
I am not sure about ansys. We had licenses and binaries for IBM power
on AIX. They are discontinuing it.... So we are looking for alternatives to
run on power7 systems. Not sure if they have support for linux x86(_64)
or if that will disapear too.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Volker Braun
2012-02-14 23:39:20 UTC
Permalink
IBM discontinuing AIX? Is this some kind of joke? IBM is more than happy to
sell Power7 systems with AIX. But if you have the hardware lying around and
no money to buy a recent AIX version then there are also plenty of linux
options. The only thing that is for sure is that you can't run Windows on
them ;-)
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Dr. David Kirkby
2012-02-14 23:48:17 UTC
Permalink
Post by Volker Braun
IBM discontinuing AIX? Is this some kind of joke? IBM is more than happy to
sell Power7 systems with AIX. But if you have the hardware lying around and
no money to buy a recent AIX version then there are also plenty of linux
options. The only thing that is for sure is that you can't run Windows on
them ;-)
Windows NT 3.51 runs on PowerPC. Not sure if that means it runs on Power 7 systems.


Dave
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
François Bissey
2012-02-15 00:26:07 UTC
Permalink
Post by Volker Braun
IBM discontinuing AIX? Is this some kind of joke? IBM is more than happy to
sell Power7 systems with AIX. But if you have the hardware lying around and
no money to buy a recent AIX version then there are also plenty of linux
options. The only thing that is for sure is that you can't run Windows on
them ;-)
Thanks for completely misleading my sentence for a few laugh.
In case it wasn't I meant ansys is discontinuing their support for power.
Now if IBM were to ditch AIX I would be a happy boy.

Didn't know Windows NT ran on powerPC, knew they had a version
for alpha (in x86 emulation). Wether it would run on power7 would
depend on what level they supported, if it ran on power4 it may
be possible but I wouldn't want to try to boot it.

Francois
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Dr. David Kirkby
2012-02-15 00:18:55 UTC
Permalink
Post by Volker Braun
Post by David Kirkby
is one such example. I was recently speaking to one of their engineers,
who said some customers had simultions taking weeks, so they would get
Ansys (the vendor) to simulate them on more powerful hardware, reducing the
time from weeks to days.
If its true that Anysys has a reasonably-sized cluster (say, 100+ nodes)
that they rent out to customers then I'm pretty sure it doesn't run on
Windows, especially if their own code is cross-platform. This is what I
meant. The fact that their code also runs on desktops doesn't make the
desktop a HPC platform.
http://www.ansys.com/ansysonwindows/ansys-performance-on-windows.pdf

says things like:


"Scale up to a cluster running Windows HPC Server 2008, and you'll be able to
consider more detailed and accurate simulations or study a range of design ideas
or operating conditions. Either way, you can leverage your existing Windows
expertise and IT infrastructure, and maximize the return on your investment in
simulation."

The University of South Florida obviously has HFSS

http://rc.usf.edu/trac/doc/wiki/HFSS

and states

"
***Batch Execution on Windows HPC***

Using Windows HPC Server 2008 R2, we can easily dispatch jobs that require more
resources than a desktop system can provide. Configuring your desktop
installation to work with the Windows HPC scheduler is a simple process."

I would have thought that is high performance computing on Windows.

I'm going on an HFSS course at the end of this month. I'll ask the engineer who
told me they rent out computer resources what OS they use for this.

Dave
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Volker Braun
2012-02-15 01:40:04 UTC
Permalink
While it would be an funny retro-computing exercise, I don't think Windows
NT 3.51 can run on Power7. It is binary compatible under certain conditions
but i'd be surprised if a whole OS can escape all caveats.

I actually know one or two Windows-based HPC clusters in academic research.
In the cases I know, they are the result of donations from Microsoft...
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Dr. David Kirkby
2012-02-15 06:13:26 UTC
Permalink
Post by Volker Braun
While it would be an funny retro-computing exercise,
No, it would be a painful one. Even if you could install it, most modern
software would not run on it.
Post by Volker Braun
I don't think Windows
NT 3.51 can run on Power7. It is binary compatible under certain conditions
but i'd be surprised if a whole OS can escape all caveats.
Perhaps. I'll leave you to try if you think it would be funny!
Post by Volker Braun
I actually know one or two Windows-based HPC clusters in academic research.
In the cases I know, they are the result of donations from Microsoft...
Ansys don't tend to support a lot of operating systems.

* On Linux they only support Redhat 4 and 5, and SUSE 10 and 11.
* On Windows they only support XP, 7 and Windows HPC Server 2008 R2.

So Windows Vista, and all the free Linux distributions like Ubuntu are not
supported. Neither is OS X.

http://www.ansys.com/Support/Platform+Support/Ansoft+Products+14.0

The fact they do support Windows HPC Server 2008 R2 makes me think they have
serious customers paying serious amounts of money to use it. HFSS is *very*
expensive. A single HFSS license for one machine is around $100 k using just one
solver. So I very much doubt one could get a commercial HFSS license for an HPC
cluster for under $500 k.

Academic licenses are much cheaper, so I doubt Ansys would support Windows HPC
Server 2008 R2 just because a few universities have HPC clusters donated by
Microsoft. I believe they must have serious commercial customers using HPC on
Windows. But I might be wrong of course.

Dave
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Volker Braun
2012-02-15 06:50:11 UTC
Permalink
Post by Dr. David Kirkby
So Windows Vista, and all the free Linux distributions like Ubuntu are not
supported. Neither is OS X.
If you have a cluster you obviously don't put Ubuntu on the nodes (there
are special cluster distributions like ROCK), so its completely
understandable that they don't actively support it. That doesn't mean that
it doesn't run on Ubuntu, they just don't actively test it.

I seems HFSS is really a desktop program that scales up to reach the bottom
of what can be considered HPC. They support Windows because thats whats
preinstalled on most desktops, not because its a good or popular clustering
solution.

Currently, one (1) of the top 500 supercomputers runs Windows
http://i.top500.org/stats.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Dima Pasechnik
2012-02-16 01:44:24 UTC
Permalink
Post by Dr. David Kirkby
Post by Volker Braun
While it would be an funny retro-computing exercise,
No, it would be a painful one. Even if you could install it, most modern
software would not run on it.
Post by Volker Braun
I don't think Windows
NT 3.51 can run on Power7. It is binary compatible under certain conditions
but i'd be surprised if a whole OS can escape all caveats.
Perhaps. I'll leave you to try if you think it would be funny!
Post by Volker Braun
I actually know one or two Windows-based HPC clusters in academic research.
In the cases I know, they are the result of donations from Microsoft...
Ansys don't tend to support a lot of operating systems.
* On Linux they only support Redhat 4 and 5, and SUSE 10 and 11.
* On Windows they only support XP, 7 and Windows HPC Server 2008 R2.
So Windows Vista, and all the free Linux distributions like Ubuntu are not
supported. Neither is OS X.
Centos is essentially a free clone of Redhat distro's.
Most everything that runs on Redhat would run on Centos without any problem.

Dima
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Dr. David Kirkby
2012-02-16 06:36:08 UTC
Permalink
Post by Dima Pasechnik
Post by Dr. David Kirkby
Post by Volker Braun
While it would be an funny retro-computing exercise,
No, it would be a painful one. Even if you could install it, most modern
software would not run on it.
Post by Volker Braun
I don't think Windows
NT 3.51 can run on Power7. It is binary compatible under certain conditions
but i'd be surprised if a whole OS can escape all caveats.
Perhaps. I'll leave you to try if you think it would be funny!
Post by Volker Braun
I actually know one or two Windows-based HPC clusters in academic research.
In the cases I know, they are the result of donations from Microsoft...
Ansys don't tend to support a lot of operating systems.
* On Linux they only support Redhat 4 and 5, and SUSE 10 and 11.
* On Windows they only support XP, 7 and Windows HPC Server 2008 R2.
So Windows Vista, and all the free Linux distributions like Ubuntu are not
supported. Neither is OS X.
Centos is essentially a free clone of Redhat distro's.
Most everything that runs on Redhat would run on Centos without any problem.
Dima
Yes, that's what I was using in fact

[***@blackcap ~]$ cat /etc/redhat-release
Red Hat Enterprise Linux AS release 4 (Nahant Update 7)
#CentOS release 4.7 (Final)

I had to edit /etc/redhat-release to fool the software to thinking it was on a
Redhat 4.7 system, when in fact it was CentOS 4.7.

Oracle Linux is also I believe a clone of Redhat - perhaps not such a close
clone as CentOS. I assume Oracle add something to it. 4

The point I was making is that if a company like Ansys decide to support Windows
HPC Server 2008 R2, whilst only supporting a very limited number of OSs, they
must believe that there is a business case for using an HPC application on Windows.

Volker Braun said:

"I seems HFSS is really a desktop program that scales up to reach the bottom of
what can be considered HPC"

I can't argue with his comments, as there's no formal definition of HPC, but I
do disagree that there is zero penetration of HPC to Windows, which was his
earlier comment.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
rjf
2012-02-16 00:32:36 UTC
Permalink
Post by Volker Braun
Post by rjf
shame on you for turning your back on what is still the
most wide-spread operating system (or family of systems) on home
and office computers.
The most common home computer is a game console. While techincally an XBox
is a windows PC, very few normal Windows programs run on it.
Also, Windows has zero market share in HPC.
You can look at the statistics for downloading of Maxima from
sourceforge.

The number of downloads on windows is about 216,000 since August.
The number of downloads on Linux is about 675.
Post by Volker Braun
primary goal should be to have a native port for gamers and secretaries
then I guess I have no choice but to drop everything else and work on the
Windows API du jour.
Whatever.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Volker Braun
2012-02-16 01:10:31 UTC
Permalink
Post by rjf
The number of downloads on windows is about 216,000 since August.
The number of downloads on Linux is about 675.
Many distributions ship with a maxima package, for example Fedora.
Microsoft is not distributing maxima with Windows. Hence only windows users
will download maxima from sourceforge.

Also, maxima is not really a HPC application.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Jason Grout
2012-02-16 03:03:16 UTC
Permalink
Post by Volker Braun
Many distributions ship with a maxima package, for example Fedora.
Microsoft is not distributing maxima with Windows. Hence only windows
users will download maxima from sourceforge.
Not to mention that the sf number doesn't count, for example, Sage
distributing maxima.

Jason
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
rjf
2012-02-16 15:32:57 UTC
Permalink
Post by Volker Braun
The number of downloads on windows is about 216,000  since August.
The number of downloads on Linux is about 675.
Many distributions ship with a maxima package, for example Fedora.
Microsoft is not distributing maxima with Windows. Hence only windows users
will download maxima from sourceforge.
It would be interesting to know how many people using some linux
distribution actually
install Maxima. (And presumably the number of Maximas exceeds the
number
of Sages, since every Sage includes Maxima).
Nevertheless, a few hundred thousand voluntarily going out and
installing
Maxima on Window is an expression of interest from Windows users that
is
hard to deny.
Post by Volker Braun
Also, maxima is not really a HPC application.
It could be a front end to such a system, but I agree it is not
itself.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Keshav Kini
2012-02-14 06:17:46 UTC
Permalink
Post by rjf
Post by Dima Pasechnik
(RJF) This is not a problem if you drop the requirement that every recipient
of Sage must be able to COMPILE stuff locally.  Just have one person
compile the stuff once and distribute dll files.
this means distributing a seriously crippled system.
How so?
Does that mean that only those people who compile Sage themselves
have a non-crippled system?
Being able to write and build your own Cython code is an important
part of normal non-trivial usage of Sage. This requires a functioning
C compiler.

-Keshav

----
Join us in #sagemath on irc.freenode.net !
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Dima Pasechnik
2012-02-14 14:10:10 UTC
Permalink
Post by Keshav Kini
Post by rjf
Post by Dima Pasechnik
(RJF) This is not a problem if you drop the requirement that every recipient
of Sage must be able to COMPILE stuff locally.  Just have one person
compile the stuff once and distribute dll files.
this means distributing a seriously crippled system.
How so?
Does that mean that only those people who compile Sage themselves
have a non-crippled system?
Being able to write and build your own Cython code is an important
part of normal non-trivial usage of Sage. This requires a functioning
C compiler.
indeed, the crippling is akin to a Lisp system that cannot compile Lisp programs,
but can only interpret them.

Dima
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Keshav Kini
2012-02-14 14:14:46 UTC
Permalink
This is a bit strange, isn't it? Shouldn't it say something like
or similar?

-Keshav

P.S. Testing sending this mail via Gnus, hopefully this works :) (shut
up about "mismatched parentheses", emacs...)
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Keshav Kini
2012-02-14 14:16:57 UTC
Permalink
Oops. Reply-To header got me there...

Sorry for the noise, still trying to wrap my head around Hardcore
Email (tm) and Hardcore Editor (tm) :)

-Keshav
Post by Keshav Kini
This is a bit strange, isn't it? Shouldn't it say something like
or similar?
-Keshav
P.S. Testing sending this mail via Gnus, hopefully this works :) (shut
up about "mismatched parentheses", emacs...)
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Jason Grout
2012-02-14 14:19:05 UTC
Permalink
Post by Keshav Kini
P.S. Testing sending this mail via Gnus, hopefully this works :) (shut
up about "mismatched parentheses", emacs...)
Thunderbird: +1.

Jason
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
David Kirkby
2012-02-14 16:19:33 UTC
Permalink
Post by William Stein
Post by Dr. David Kirkby
Post by rjf
Microsoft and Google/Motorola are suing each other.
Do you get money from Microsoft?
Just a thought.
You could propose to Google to port Sage to run on Windows/ natively,
not.
I believe a complete native port would be an almost impossibility. I
don't think
Post by Dr. David Kirkby
many developers are keen on Windoze, so it would be less easy to attract
developers unless they were paid to do it.
So ask for money to pay them not to do it, from Google.
Frankly, I think a person competent in Windows technology could do
this.
In my own experience, Maxima runs fine under windows.
Also in my experience, GMP and MPFR run under Windows, but maybe
require additional non-free software development environments.
This is not a problem if you drop the requirement that every recipient
of Sage must be able to COMPILE stuff locally. Just have one person
compile the stuff once and distribute dll files.
I suppose
it is possible that some Sage propeller-heads have written python or C
code that cannot be compiled under windows, but it seems to me more
likely that it is a question of finding a skilled person, and not so
much of problematical code. A skilled person who cares to look at
code base, and has a spare week or two.
I've written a fair amount of software over the years - some Windows
applications with a GUI, but most are Unix applications. I think I know a
thing or two about porting software.

IMHO, a native port of Sage to Windows could not be done in a week or two.
Perhaps a Cygwin port could, but I'm talking of a native port, where the
code runs directly on Windows, without any Linux virtual machines,
emulators or similar.

If you wish, I'll offer you a 20:1 bet.. If you, or someone you know, can
get a full port of Sage done inside a month, I'll pay you $2000. On the
other hand, it you can't provide a full port, you pay me $100. William can
be the judge of whether the resulting code is a complete native port of
Sage to Windows. Let me know if you want that bet.

PS. It's interesting that Wolfram Research appear to use ATLAS in
Mathematica - at least on Solaris. I base this on the fact there is a file
'libatlas.so', or some similar name in the Mathematica distribution. I
personally think ATLAS would be one of the more difficult programs to port
to Windows. I've often pondered if Wolfram Research have ported ATLAS to
Windows, or if they use a different library on Windows. The latter would
seem odd (why maintain two lots of code base?), so it suggests to me they
have probably ported ATLAS to Windows. But that would seem very hard - to
do it properly anyway.

Dave
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Volker Braun
2012-02-14 23:07:19 UTC
Permalink
Post by David Kirkby
PS. It's interesting that Wolfram Research appear to use ATLAS in
Mathematica - at least on Solaris. I base this on the fact there is a file
'libatlas.so', or some similar name in the Mathematica distribution.
An internet serach suggests that ATLAS can in fact be compiled with cygwin
and even MinGW on windows. ATLAS certainly doesn't have any need for
fork(), and even compiles its own build tools.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
rjf
2012-02-16 00:22:29 UTC
Permalink
Post by David Kirkby
IMHO, a native port of Sage to Windows could not be done in a week or two.
Perhaps a Cygwin port could, but I'm talking of a native port, where the
code runs directly on Windows, without any Linux virtual machines,
emulators or similar.
I see no reason to reject MinGW, Cygwin, or other libraries as part of
a Sage system on Windows.
Post by David Kirkby
If you wish, I'll offer you a 20:1 bet.. If you, or someone you know, can
get a full port of Sage done inside a month, I'll pay you $2000.
I think that a month (160 hours X expert rate of, say, $500/hour)
would
do it. That is far more than $2,000.

There may also be some stupid-ass code in some of the random
facilities
in Sage that would have to be either understood or written around
because
the writer essentially dared anyone to do (whatever) without Unix.
A dumb move, but not impossible. In any case, I'm not a betting man,
other than retirement funds..


On the
Post by David Kirkby
other hand, it you can't provide a full port, you pay me $100. William can
be the judge of whether the resulting code is a complete native port of
Sage to Windows. Let me know if you want that bet.
Hm, that sounds like an unbiased judge. :)
Post by David Kirkby
PS. It's interesting that Wolfram Research appear to use ATLAS in
Mathematica - at least on Solaris. I base this on the fact there is a file
'libatlas.so', or some similar name in the Mathematica distribution. I
personally think ATLAS would be one of the more difficult programs to port
to Windows. I've often pondered if Wolfram Research have ported ATLAS to
Windows, or if they use a different library on Windows. The latter would
seem odd (why maintain two lots of code base?), so it suggests to me they
have probably ported ATLAS to Windows. But that would seem very hard - to
do it properly anyway.
Dave
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
kcrisman
2012-02-16 02:31:11 UTC
Permalink
Post by rjf
Post by David Kirkby
IMHO, a native port of Sage to Windows could not be done in a week or two.
Perhaps a Cygwin port could, but I'm talking of a native port, where the
code runs directly on Windows, without any Linux virtual machines,
emulators or similar.
I see no reason to reject MinGW, Cygwin, or other libraries as part of
a Sage system on Windows.
Then you haven't tried to actually do this, which of course we know.
Unfortunately, even getting Maxima to work right on Cygwin with ECL
was nontrivial lately because of how Juanjo had to do forking (which
he's since gotten around entirely) which does not work on Cygwin
properly. (And I say this out of sheer experience trying to get it to
work, not because I know anything about forking. Those who do are
even more emphatic about it.)
Post by rjf
If you wish, I'll offer you a 20:1 bet.. If you, or someone you know, can
Post by David Kirkby
get a full port of Sage done inside a month, I'll pay you $2000.
I think that a month (160 hours X expert rate of, say, $500/hour)
would
do it.  That is far more than $2,000.
$500/hr? NSF's rate on MAA grants is $50/hr! I guess you should all
just ditch academia and work for a few months, cash in six figures,
and call it a year so you can work on open source software the rest of
the year :)

Don't need a judge, either; Sage passes all doctests and works on
Windows without virtualization or the *Gw*s, there you go.

Anyway, all this is hypothetical, talk of bets or whatever. There
exist people (just a few) who could actually do the Cygwin port
quickly if they worked on it *full time*, but they have judged their
time better spent elsewhere, and I don't blame them. And the MSVC
(sp?) port would be a few orders of magnitude worse. Yes, there is
probably a lot of C code in some of the packages in Sage which are not
really up to standards or are heavily reliant on Gnu-specific stuff...
Dave K. has been very helpful along these lines.

But that's what we need to include in order to get the highest-quality
*mathematics* in Sage. And that, in the end, is the point. The
notebook and Sage cell make it less and less necessary to do anything
outside the cloud anyway.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
rjf
2012-02-16 15:43:11 UTC
Permalink
Post by kcrisman
Post by rjf
Post by David Kirkby
IMHO, a native port of Sage to Windows could not be done in a week or two.
Perhaps a Cygwin port could, but I'm talking of a native port, where the
code runs directly on Windows, without any Linux virtual machines,
emulators or similar.
I see no reason to reject MinGW, Cygwin, or other libraries as part of
a Sage system on Windows.
Then you haven't tried to actually do this, which of course we know.
Unfortunately, even getting Maxima to work right on Cygwin with ECL
... maybe because running ECL is the wrong choice, dictated by a
misguided policy about what software is politically acceptable.
Post by kcrisman
was nontrivial lately because of how Juanjo had to do forking (which
he's since gotten around entirely) which does not work on Cygwin
properly.  (And I say this out of sheer experience trying to get it to
work, not because I know anything about forking.  Those who do are
even more emphatic about it.)
 > If you wish, I'll offer you a 20:1 bet.. If you, or someone you
know, can
Post by rjf
Post by David Kirkby
get a full port of Sage done inside a month, I'll pay you $2000.
I think that a month (160 hours X expert rate of, say, $500/hour)
would
do it.  That is far more than $2,000.
$500/hr? NSF's rate on MAA grants is $50/hr!
I suppose the NSF doesn't have any refrigerators, dishwashers, or
plumbing.
Try to get a repair person to your house at $50 an hour. Although it
is easy
to get a PhD mathematician to your house. Just order a pizza.
Post by kcrisman
I guess you should all
just ditch academia and work for a few months, cash in six figures,
and call it a year so you can work on open source software the rest of
the year :)
I said an expert on windows, not open source software.
Post by kcrisman
Don't need a judge, either; Sage passes all doctests and works on
Windows without virtualization or the *Gw*s, there you go.
Anyway, all this is hypothetical, talk of bets or whatever.  There
exist people (just a few) who could actually do the Cygwin port
quickly if they worked on it *full time*, but they have judged their
time better spent elsewhere, and I don't blame them.
So we agree.

 And the MSVC
Post by kcrisman
(sp?) port would be a few orders of magnitude worse.
I doubt it. If there really are bad C compilers, that might be a
different
situation.

My own personal experience has been that I will struggle for several
days
on some such project, and then get help from a true expert who solves
the problem in 60 seconds.


 Yes, there is
Post by kcrisman
probably a lot of C code in some of the packages in Sage which are not
really up to standards or are heavily reliant on Gnu-specific stuff...
Dave K. has been very helpful along these lines.
But that's what we need to include in order to get the highest-quality
*mathematics* in Sage.  And that, in the end, is the point.  The
notebook and Sage cell make it less and less necessary to do anything
outside the cloud anyway.
My own feeling is that pieces of Sage are just not necessarily high
quality. Consider all the code written in Python. Is it really the
highest performance code, or is just the stuff that happens to be
in Python?

I could be more specific about the pieces I have looked at, but to be
accurate I'd have to look at them again to make sure they hadn't been
improved.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
kcrisman
2012-02-16 16:54:01 UTC
Permalink
Post by rjf
Post by kcrisman
Then you haven't tried to actually do this, which of course we know.
Unfortunately, even getting Maxima to work right on Cygwin with ECL
... maybe because running ECL is the wrong choice, dictated by a
misguided policy about what software is politically acceptable.
Perhaps; I have no horse in that race.
Post by rjf
My own personal experience has been that I will struggle for several
days
on some such project, and then get help from a true expert who solves
the problem in 60 seconds.
Correct. Unfortunately, in this case you really need someone who is
an expert in a very, very small (though nonempty) Venn diagram
intersection. There probably aren't too many folks who get Windows in
the way needed *and* are experts about internals of things like PARI
and GAP.
Post by rjf
 Yes, there is
Post by kcrisman
probably a lot of C code in some of the packages in Sage which are not
really up to standards or are heavily reliant on Gnu-specific stuff...
Dave K. has been very helpful along these lines.
But that's what we need to include in order to get the highest-quality
*mathematics* in Sage.  And that, in the end, is the point.  The
notebook and Sage cell make it less and less necessary to do anything
outside the cloud anyway.
My own feeling is that pieces of Sage are just not necessarily high
quality.  Consider all the code written in Python.  Is it really the
highest performance code, or is just the stuff that happens to be
in Python?
Mathematically, not performance. Some of it is probably quite slow.
Here the communities depart ways based on discipline, I think.

As an analogy, R, for instance, is universally complained about in
this regard. But it has so much good functionality in user-
contributed packages, that people still use it (and companies,
apparently profitable ones, create add-ons to speed it up for such
users). It's also extremely widely used, and growing all the time -
in production environments, too, not just (or even mainly) academia.
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Dima Pasechnik
2012-02-16 17:45:17 UTC
Permalink
Post by rjf
Post by kcrisman
Post by rjf
Post by David Kirkby
IMHO, a native port of Sage to Windows could not be done in a week or two.
Perhaps a Cygwin port could, but I'm talking of a native port, where the
code runs directly on Windows, without any Linux virtual machines,
emulators or similar.
I see no reason to reject MinGW, Cygwin, or other libraries as part of
a Sage system on Windows.
Then you haven't tried to actually do this, which of course we know.
Unfortunately, even getting Maxima to work right on Cygwin with ECL
... maybe because running ECL is the wrong choice, dictated by a
misguided policy about what software is politically acceptable.
Post by kcrisman
was nontrivial lately because of how Juanjo had to do forking (which
he's since gotten around entirely) which does not work on Cygwin
properly.  (And I say this out of sheer experience trying to get it to
work, not because I know anything about forking.  Those who do are
even more emphatic about it.)
 > If you wish, I'll offer you a 20:1 bet.. If you, or someone you
know, can
Post by rjf
Post by David Kirkby
get a full port of Sage done inside a month, I'll pay you $2000.
I think that a month (160 hours X expert rate of, say, $500/hour)
would
do it.  That is far more than $2,000.
By far the quickest way to make Sage run on Windows is to get an expert
to fix Cygwin's fork implementation.

The problem is that while such persons exist, they all have
signed an NDA, which would prevent them to do what's needed.

Anything esle is a huge waste of man-hours.
(of course this makes your 500$/h experts in wheel reinvention
laughing all the way to the bank...)

Frankly, that's disturbing, isn't it?

Dima
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
William Stein
2012-02-16 18:05:23 UTC
Permalink
Post by Dima Pasechnik
Post by rjf
Post by kcrisman
Post by rjf
Post by David Kirkby
IMHO, a native port of Sage to Windows could not be done in a week or two.
Perhaps a Cygwin port could, but I'm talking of a native port, where the
code runs directly on Windows, without any Linux virtual machines,
emulators or similar.
I see no reason to reject MinGW, Cygwin, or other libraries as part of
a Sage system on Windows.
Then you haven't tried to actually do this, which of course we know.
Unfortunately, even getting Maxima to work right on Cygwin with ECL
... maybe because running ECL is the wrong choice, dictated by a
misguided policy about what software is politically acceptable.
Post by kcrisman
was nontrivial lately because of how Juanjo had to do forking (which
he's since gotten around entirely) which does not work on Cygwin
properly.  (And I say this out of sheer experience trying to get it to
work, not because I know anything about forking.  Those who do are
even more emphatic about it.)
 > If you wish, I'll offer you a 20:1 bet.. If you, or someone you
know, can
Post by rjf
Post by David Kirkby
get a full port of Sage done inside a month, I'll pay you $2000.
I think that a month (160 hours X expert rate of, say, $500/hour)
would
do it.  That is far more than $2,000.
By far the quickest way to make Sage run on Windows is to get an expert
to fix Cygwin's fork implementation.
The problem is that while such persons exist, they all have
signed an NDA, which would prevent them to do what's needed.
Why do you think it's *possible* to have a good fork implementation in
Cygwin? I had the impression that it is impossible.

Also, there is much, much more wrong with Cygwin than just fork...
Post by Dima Pasechnik
Anything esle is a huge waste of man-hours.
(of course this makes your 500$/h experts in wheel reinvention
laughing all the way to the bank...)
Frankly, that's disturbing, isn't it?
Dima
--
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
--
William Stein
Professor of Mathematics
University of Washington
http://wstein.org
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Dima Pasechnik
2012-02-16 18:14:15 UTC
Permalink
Post by William Stein
Post by Dima Pasechnik
Post by rjf
Post by kcrisman
Post by rjf
Post by David Kirkby
IMHO, a native port of Sage to Windows could not be done in a week or two.
Perhaps a Cygwin port could, but I'm talking of a native port, where the
code runs directly on Windows, without any Linux virtual machines,
emulators or similar.
I see no reason to reject MinGW, Cygwin, or other libraries as part of
a Sage system on Windows.
Then you haven't tried to actually do this, which of course we know.
Unfortunately, even getting Maxima to work right on Cygwin with ECL
... maybe because running ECL is the wrong choice, dictated by a
misguided policy about what software is politically acceptable.
Post by kcrisman
was nontrivial lately because of how Juanjo had to do forking (which
he's since gotten around entirely) which does not work on Cygwin
properly.  (And I say this out of sheer experience trying to get it to
work, not because I know anything about forking.  Those who do are
even more emphatic about it.)
 > If you wish, I'll offer you a 20:1 bet.. If you, or someone you
know, can
Post by rjf
Post by David Kirkby
get a full port of Sage done inside a month, I'll pay you $2000.
I think that a month (160 hours X expert rate of, say, $500/hour)
would
do it.  That is far more than $2,000.
By far the quickest way to make Sage run on Windows is to get an expert
to fix Cygwin's fork implementation.
The problem is that while such persons exist, they all have
signed an NDA, which would prevent them to do what's needed.
Why do you think it's *possible* to have a good fork implementation in
Cygwin? I had the impression that it is impossible
Quoting http://comments.gmane.org/gmane.comp.python.cython.devel/8730:

% Cygwin's problem is that hooking up a process created by
% ZwCreateProcess to the Win32 or SUA subsystem is undocumented; neither
% MS documentation or Nebbet cover that. That is why Cygwin does not
% implement a copy-on-write fork yet, although most modern hardware
% supports it.
Post by William Stein
Also, there is much, much more wrong with Cygwin than just fork...
aside of absense of a 64-bit version (and the dreadful fork crap),
it's workable...

Dima
Post by William Stein
Post by Dima Pasechnik
Anything esle is a huge waste of man-hours.
(of course this makes your 500$/h experts in wheel reinvention
laughing all the way to the bank...)
Frankly, that's disturbing, isn't it?
Dima
--
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
--
William Stein
Professor of Mathematics
University of Washington
http://wstein.org
--
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Dr. David Kirkby
2012-02-18 11:50:53 UTC
Permalink
Post by rjf
Post by David Kirkby
IMHO, a native port of Sage to Windows could not be done in a week or two.
Perhaps a Cygwin port could, but I'm talking of a native port, where the
code runs directly on Windows, without any Linux virtual machines,
emulators or similar.
I see no reason to reject MinGW, Cygwin, or other libraries as part of
a Sage system on Windows.
Post by David Kirkby
If you wish, I'll offer you a 20:1 bet.. If you, or someone you know, can
get a full port of Sage done inside a month, I'll pay you $2000.
I think that a month (160 hours X expert rate of, say, $500/hour)
would
do it. That is far more than $2,000.
Yes, it's a ridiculously high hourly rate too for a programmer. In the current
economic climate, a contractor would be reasonably happy earning $500/day, not
an hour.

I think $2000 is quite reasonable for what you considered was 1-2 weeks work,
although I was giving you a month.


You are either stupid (which I don't think you are), just difficult for the sake
of it (which I think you are), or grossly underestimate the amount of work
required.

Dave
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
rjf
2012-02-18 18:54:50 UTC
Permalink
Post by Dr. David Kirkby
Post by rjf
Post by David Kirkby
IMHO, a native port of Sage to Windows could not be done in a week or two.
Perhaps a Cygwin port could, but I'm talking of a native port, where the
code runs directly on Windows, without any Linux virtual machines,
emulators or similar.
I see no reason to reject MinGW, Cygwin, or other libraries as part of
a Sage system on Windows.
Post by David Kirkby
If you wish, I'll offer you a 20:1 bet.. If you, or someone you know, can
get a full port of Sage done inside a month, I'll pay you $2000.
I think that a month (160 hours X expert rate of, say, $500/hour)
would
do it.  That is far more than $2,000.
Yes, it's a ridiculously high hourly rate too for a programmer. In the current
economic climate, a contractor would be reasonably happy earning $500/day, not
an hour.
I think $2000 is quite reasonable for what you considered was 1-2 weeks work,
although I was giving you a month.
If you care to look around at what (say) lawyers make,
or consultants who really understand commercial software you would
understand
that $500 an hour -- for the right person -- is possible.
Post by Dr. David Kirkby
You are either stupid (which I don't think you are), just difficult for the sake
of it (which I think you are), or grossly underestimate the amount of work
required.
Could be any of those, or I could be right.

RJF
--
To post to this group, send an email to sage-***@googlegroups.com
To unsubscribe from this group, send an email to sage-devel+***@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
Continue reading on narkive:
Loading...