Let lrcalc build as a shared library on Cygwin.

[jpflori]

We need to pass the -no-undefined flag to libtool to build a shared library on Cygwin.

This was added to the 1.1.6 beta spkg distributed within Sage.

Following discussion with upstream it was decided that upstream would distribute an alternate autotools-based lrcalc, named lrcalc-sage, using the spkg modified sources.

The sources have been published upstream: ​http://www.math.rutgers.edu/~asbuch/lrcalc/

Therefore we stop tracking the src directory (which was not such a good idea) as it will be done upstream. (And as a consequence the diff attached here is quite big but the spkg does not grows.)

Try spkg at ​http://boxen.math.washington.edu/home/jpflori/lrcalc-1.1.6.spkg

[jpflori]

Spkg diff, for review only.

[jpflori]

New spkg uploaded.

I've rerun autoreconf with recent versions of the tool as the current state of the build system looked messy (and trying to use Jeroen smart spkg to produce a minimal diff complained a little and produced something broken).

It works fine on Linux, I won't be able to check on Cygwin before this night or tomorrow.

[jpflori]

Upstream is ready to distribute an alternative official package using autotools, in addition to the original one using a custom build system.

So I propose to wait for everybody to be happy with the spkg here, send it upstream, repackage it "properly" here and merge it.

[nthiery]

I checked the changes in the diff, and am happy with them. Feel free to set a positive review on my behalf (I'll be offline the next couple days).

[jpflori]

Just a confirmation that this works as expected on Cygwin.

After our discussion with upstream, I will remove tracking of the src directory from the spkg and repost a new spkg. The src directory will now be tracked in a separate hg tree kept upstream.

[jpflori]

Spkg diff, for review only.

[kcrisman]

Wait, did this not work right? Or is this just an enhancement? (I know there is a parent ticket, but I forget the number.)

[jpflori]

It was not build as shared before. I cannot say that the static archive produced was dysfunctional, I've not run the tests, so we can consider it an enhancement if not a defect.

[jpflori]

There have been some minimal changes to the tarball posted upstream so we need to resync.

[jpflori]

I've resynced the src folder with the upstream tarball (mostly copyright additions and some autotools regenration stuff) and polished a little bit SPKG.txt.

Nicolas could you have a quick look at the spkg and put this to positive review as we planned a few monthes ago?

[jpflori]

Spkg diff, for review only.

[dimpase]

Replying to jpflori:

I've resynced the src folder with the upstream tarball (mostly copyright additions and some autotools regenration stuff) and polished a little bit SPKG.txt.

Nicolas could you have a quick look at the spkg and put this to positive review as we planned a few monthes ago?

It builds on cygwin, but fails tests. More precisely the following fails:

sage: from sage.libs.lcalc.lcalc_Lfunction import *
---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
<ipython-input-1-015a4f363123> in <module>()
----> 1 from sage.libs.lcalc.lcalc_Lfunction import *

ImportError: No such file or directory
sage:

Is it just me?

[jpflori]

Didn't you mess lcalc and lrcalc together?

[dimpase]

Replying to jpflori:

Didn't you mess lcalc and lrcalc together?

oops, indeed. lrcalc works. Positive review.