update 4ti2 to version 1.6

[dimpase]

version 1.6 is released.

Intsall the updated spkg and apply attachment: 42.patch

CC: @novoselt @miguelmarco @sagetrac-dperkinson @sagetrac-mhampton

Component: packages: optional

Keywords: 4ti2

Author: Dmitrii Pasechnik

Reviewer: Volker Braun

Merged: sage-5.13.beta1

Issue created by migration from https://trac.sagemath.org/ticket/15172

[dimpase]

comment:1

at the moment there is one test failure, which only occurs from sage -t run, but I can't reproduce it interactively:

$ ../../sage -t --optional='sage,4ti2' sage/interfaces/four_ti_2.py
Running doctests with ID 2013-09-07-09-09-54-b455bb50.
Doctesting 1 file.
sage -t sage/interfaces/four_ti_2.py
**********************************************************************
File "sage/interfaces/four_ti_2.py", line 284, in sage.interfaces.four_ti_2.FourTi2.call
Failed example:
    four_ti_2.read_matrix("test_file.gro") # optional - 4ti2
Expected:
    [-5  0  2]
    [-5  3  0]
Got:
    []
**********************************************************************
1 item had failures:
   1 of   5 in sage.interfaces.four_ti_2.FourTi2.call
    [50 tests, 1 failure, 0.55 s]
----------------------------------------------------------------------
sage -t sage/interfaces/four_ti_2.py  # 1 doctest failed

[jdemeyer]

comment:4

Why add # optional - 4ti2 where it was not needed before?

[jdemeyer]

comment:5

The following

        cwd = os.getcwd() # Save working directory in order to restore it.
        os.chdir(self.directory())

        subprocess.call(cmd, shell=True)

        os.chdir(cwd) # Restore previous working directory.

should be replaced by

        subprocess.call(cmd, shell=True, cwd=self.directory())

[novoselt]

comment:6

As I understand in the patch "-q" is for quiet mode, but why there is a need to silence stderr explicitly? Shouldn't commands output no errors if used correctly?

[dimpase]

comment:7

Replying to @novoselt:

As I understand in the patch "-q" is for quiet mode, but why there is a need to silence stderr explicitly? Shouldn't commands output no errors if used correctly?

there is no '-q' switch for ppi, for some reason. So its stderr output needs to be silenced.

[dimpase]

comment:8

Replying to @jdemeyer:

Do you mean it's an equivalent way to do the same thing?

OK, I'll change this. Meanwhile I just uploaded patch which fixes the issue that was stopping me here; it's actually due to filename semantics changed. Now groebner bla.foo appends the suffix gro to bla.foo, rather than replacing the suffix.

[dimpase]

Changed work issues from tmp_dir related doctest issue to none

[dimpase]

Attachment: 42.patch.gz

updated patch

[dimpase]

comment:9

OK, it looks as if I fixed everything.

[dimpase]

comment:10

There are quite a few changes in the 4ti2-dependent part of sandpile.py, due to a change in output format of zslove, which prints matrix dimensions on one row now, and changed order in its output. Also, I needed to take care of deprecated vertex_set in SimplicialComplex.

[jdemeyer]

comment:11

Replying to @dimpase:

Replying to @jdemeyer:

Do you mean it's an equivalent way to do the same thing?

Essentially yes. Instead of changing directory globally and then changing it back, we simply change directory for the executed command.

[dimpase]

comment:12

I've made a small change in spkg-install. Now everything, not only executables, is installed, so that one can link against 4ti2's libraries.

[vbraun]

Reviewer: Volker Braun

[vbraun]

comment:14

Looks good to me.

[vbraun]

Author: Dmitrii Pasechnik

[jdemeyer]

Merged: sage-5.13.beta1