Latest Cygwin status

[kcrisman]

Just forwarding from sage-devel to apprise folks of the most recent developments.

 

Hi JP,

I would love to try this on an old laptop I've been loaned for this purpose (running XP!) but I'm afraid that I don't have the time to follow through your lovingly detailed comments on the Trac wiki for this :)  Is there any way you can create a tarball with all the "right" spkgs that one could try to use just for testing?

Most patches have been integrated into Sage, so you merely need to download to the updated spkg for:
* eclib at #13325
* mpir at #12115
* ntl at #11635
* ecl at #13324
and the following ones which will hopefully be integrated quite soon:
* pari at #13333
* python at #13319
This should let Sage build (in fact IIRC eclib and ntl are not actually needed to build Sage), although you'll have to rebase your dlls from time to time (and I fear there is nothing to do against that) and restart the build.
There will be problems left, see e.g. #9167 for further problems with ECL, #13350 and #13351 for other simple fixes, and #13354 for further autotools fixes.

I've not touched my installation for a few months so I don't really feel to package anything now.
Hopefully I'll have more time before the end of the year, and with some help from reviewers the situation should get much better.

And reports from XP would be really nice as I only test on 7. (I have an old 32 bits laptop with an XP installation but the screen is broken)

[Dima Pasechnik]

It seems that Sage 5.7.beta0 is almost there in sense of building and
running Sage on a recent Cygwin (on a 64-bit Windows 7 system).

I've built it with very minimal hacks (a couple of trac tickets which
didn't make it into this release yet) and rebasing twice during the
installation.
Here is the list, with brief/incomplete diagnosis/symptoms,
of failed tests during "make test". (full log here:
http://boxen.math.washington.edu/home/dima/tmp/cygwin_sage-5.7.beta0.test.log)

Out of 21 failed tests:
3 are numerical noise.
2 are due to segfaults in the external program "nef"
4 are due to not implemented memory usage function
5 are partly due to a problem converting Python long to C long
(probably 32/64-bit issue); in Singular and in sparse matrix
2 are due an interpolation error in gsl
2 are due to a Singular problem, like
"Exception KeyError: (The ring pointer -0x1bbd6b8,) "

the rest are things like inline fortran and cython problems, probably
not hard to fix.

---------------------------------------------------------
1. "devel/sage/sage/algebras/free_algebra.py"
same as in other Singular/plural related errors below

2. "devel/sage/sage/categories/category_singleton.pyx"
sage: hash(Rings()) == id(Rings) # indirect doctest
Expected:
True
Got:
False

3. "devel/sage/sage/functions/other.py"
numerical noise

4. "devel/sage/sage/geometry/lattice_polytope.py"
segfaults in "nef" --- see below

5. "devel/sage/sage/graphs/genus.pyx"
memory usage not implemented on platform cygwin_nt-6.1-wow64

6. "devel/sage/sage/gsl/ode.pyx"
gsl: interp.c:150: ERROR: interpolation error

7. "devel/sage/sage/libs/lcalc/lcalc_Lfunction.pyx"
library import problem - did we see this with ECL too?
sage: from sage.libs.lcalc.lcalc_Lfunction import *
ImportError: No such file or directory

8. "devel/sage/sage/libs/pari/gen.pyx"
memory usage not implemented on platform cygwin_nt-6.1-wow64

9. "devel/sage/sage/libs/singular/function.pyx"
Exception KeyError: (The ring pointer -0x1c2f77c,) in
'sage.libs.singular.ring.singular_ring_delete' ignored
and
(probably C long is too short on Cygwin)
OverflowError: Python int too large to convert to C long

10. "devel/sage/sage/libs/singular/groebner_strategy.pyx"
(probably C long is too short on Cygwin)
OverflowError: Python int too large to convert to C long

11. "devel/sage/sage/matrix/matrix_sparse.pyx"
(probably C long is too short on Cygwin)
OverflowError: Python int too large to convert to C long

12. "devel/sage/sage/misc/cython.py"
(is it making a .dll here?)
Compiling hello.spyx...
cp: cannot stat
`/home/dima/.sage/temp/SPMS_DIMA_W7_64/5508/spyx/hello/hello.so': No
such file or directory

13. "devel/sage/sage/misc/getusage.py"
NotImplementedError: top not implemented on platform
cygwin_nt-6.1-wow64

14. "devel/sage/sage/misc/inline_fortran.py"
probably, a misconfiguration due to an external fortran compiler used:

RuntimeError: failed to compile Fortran code:
Found executable
/usr/local/src/sage/sage-5.7.beta0/local/bin/sage-inline-fortran
gnu: no Fortran 90 compiler found
Found executable /usr/bin/ld
Found executable /usr/bin/ar
Found executable /usr/bin/ranlib
gnu: no Fortran 90 compiler found
f951: warning: -fPIC ignored for target (all code is position
independent)

15 "devel/sage/sage/plot/plot.py"
gsl: interp.c:150: ERROR: interpolation error
(no idea...)

16. "devel/sage/sage/rings/arith.py"
numerical noise

17. "devel/sage/sage/rings/polynomial/multi_polynomial_ideal.py" Singular
interface problem, messages like "OverflowError: Python int too large to
convert to C long" --- does this indicate that the platform is
misidentfied as x86_64?
and a weird one:
Exception KeyError: (The ring pointer -0x1bbd6b8,) in
'sage.libs.singular.ring.singular_ring_delete' ignored

18. "devel/sage/sage/rings/polynomial/plural.pyx"
interface problem, messages like "OverflowError: Python int too large to
convert to C long" --- does this indicate that the platform is misidentfied
as x86_64?

19. "devel/sage/sage/rings/tests.py" not implemented memory usage on Cygwin
NotImplementedError('memory usage not implemented on platform
cygwin_nt-6.1-wow64',)

20. "devel/sage/sage/schemes/toric/fano_variety.py" segfault in "nef":
RuntimeError: Error executing "nef.x -f -N -V -D -P -p" for a
polytope sequence!
Output:
/bin/sh: line 1: 4628 Segmentation fault (core dumped) nef.x
-f -N -V -D -P -p <
/home/dima/.sage/temp/SPMS_DIMA_W7_64/6116/tmp_V_s5P7 >
/home/dima/.sage/temp/SPMS_DIMA_W7_64/6116/tmp
_LHvtCc

21. "devel/sage/sage/symbolic/expression.pyx"
numerical noise

[Dima Pasechnik]

On 2013-01-27, Dima Pasechnik <dim...@gmail.com> wrote:
> It seems that Sage 5.7.beta0 is almost there in sense of building and
> running Sage on a recent Cygwin (on a 64-bit Windows 7 system).
>
> I've built it with very minimal hacks (a couple of trac tickets which
> didn't make it into this release yet) and rebasing twice during the
> installation.
> Here is the list, with brief/incomplete diagnosis/symptoms,
> of failed tests during "make test". (full log here:
> http://boxen.math.washington.edu/home/dima/tmp/cygwin_sage-5.7.beta0.test.log)
>
> Out of 21 failed tests:
> 3 are numerical noise.
> 2 are due to segfaults in the external program "nef"

this is fixed in #13960

> 4 are due to not implemented memory usage function

this is fixed in #9170

[kcrisman]

After 281 comments, http://trac.sagemath.org/sage_trac/ticket/6743 has been closed by the release manager - hooray!  I think that at some point JP uploaded a binary for download... now maybe someone can tell us how to make the ultimate goal: a one-click Cygwin-installer-cum-binary that we apparently used to have long ago - that is, a file which bundled Cygwin and the Sage Cygwin binary all in one package, no additional downloads etc. required!