Good news on the Windows/Cygwin front

[Erik Bray]

Hi all,

Here is the result of my latest run of `./sage -t -a` on Cygwin 64-bit:

$ ./sage -t -a
Running doctests with ID 2016-08-19-18-59-29-d5e4434f.
Git branch: cygwin
Using --optional=mpir,python2,sage
Doctesting entire Sage library.
Doctesting 3444 files.
...
----------------------------------------------------------------------
sage -t --warn-long 68.0 src/sage/interfaces/expect.py # 1 doctest failed
----------------------------------------------------------------------
Total time for all tests: 11840.1 seconds
cpu time: 8006.9 seconds
cumulative wall time: 10877.7 seconds


This is on a branch based on a somewhat old (maybe about a month)
version of the develop branch, plus all the fixes I've made so far for
Cygwin (not all of which have tickets yet).

The one failure is something of a fluke--I've actually *never* seen
that particular failure before this test run, and when I test that
module by itself it's passing. Nevertheless it does bother me a
little bit and I want to try to reproduce the failure. But it comes
as no surprise that there could still be semi-non-deterministic errors
in the pexpect interface.

Other than that, most things are working well. I believe there are
still some --long tests that are failing, which I have not looked into
yet. But I don't think there are many.

Once I've confirmed that I can get all the tests (including the expect
one) running reliably my next major goal is to get a build-bot set up
for Cygwin. This may be a bit challenging due to the unique
challenges of building Sage on Cygwin (namely, DLL rebasing), but not
impossible.

For further status reports on Windows I will be posting just to the
sage-windows group, so you can join that if you're interested in this
effort and ignore otherwise:
https://groups.google.com/forum/#!forum/sage-windows

There are also some tickets needing review for various Cygwin-related issues:

https://trac.sagemath.org/query?status=needs_review&status=new&component=porting%3A+Cygwin&col=id&col=summary&col=status&col=type&col=priority&col=milestone&col=component&order=priority

Thanks,
Erik

[Erik Bray]

On Wed, Aug 31, 2016 at 6:14 PM, Travis Scrimshaw <tsc...@ucdavis.edu> wrote:
> I was able to get Cygwin64 to build successfully using 7.4.beta2 + Pynac
> 0.6.9 upgrade (#21369) + interpreter hack (#19868). I did have an issue with
> twisted not picking up zope_interface, but I was able to fix it by forcing a
> reinstall of zope_interface and then building twisted again.
>
> For Cygwin32, I'm stuck at pycrypto (#21236).
>
> Best,
> Travis

Thanks for giving it a try! I'm surprised actually that it worked,
since I still have a few patches that I haven't posted yet (though I
think those have more to do with getting some tests to pass, than
actually building).

I have a nicer patch for the autogen.interpreters issue, but it
depends on https://trac.sagemath.org/ticket/20730 being accepted
first.

I do hope to look into getting Cygwin32 builds working eventually too,
and will look at the issue with pycrypto if I get a chance.

Thanks,
Erik