sh: kpsewhich: not found - Sage 4.3.4.alpha0 on Solaris

[sagetrac-drkirkby]

== Background ==
After downloading the 4.3.4.alpha0.tar I applied some patches necessary to get this to build on Solaris 10 (SPARC). Namely:

=== Patches installed to allow Sage to build properly ===

• GCC reports incorrect flags compiling descent_two_isogeny.c on Solaris 10 #7867 A patch for Python which allows Python modules to be built properly.
• Removal of pyprocessing causing problems as _multiprocessing not building on Solaris #8440 A patch for Python to allow the _multiprocessing module to build.

=== Patches installed to allow most all doctests to pass in 4.3.3 (a few fail in 4.3.4.alpha0 ===

• Numerical noise in devel/sage/sage/symbolic/constants_c.pyx #8374 Numerical noise in sage/symbolic/constants_c.pyx
• Numerical noise in devel/sage/sage/symbolic/pynac.pyx #8375 Numerical noise in sage/symbolic/pynac.pyx
• Change 'top' to 'prstat' on Solaris for get_memory_usage() and top() #8391 Change 'top' to 'prstat' on Solaris, othewise lots of doctests time out.
• Update sqlite to 3.6.22 (the latest version) #8408 Update sqlite to the latest version (otherwise doctest devel/sage/sage/graphs/graph_list.py fails on Solaris 10 (SPARC) #8397, doctest devel/sage/sage/graphs/generic_graph.py fails on Solaris 10 (SPARC) #8398, doctest devel/sage/sage/graphs/graph.py fails on Solaris 10 (SPARC) #8399, doctest devel/sage/sage/databases/database.py fails on Solaris 10 (SPARC) #8400 and doctest devel/sage/sage/graphs/graph_database.py fails on Solaris 10 (SPARC) #8401 all fail).

== The problems ==

Running the long doctests I see:

sage -t  -long "devel/sage/sage/categories/finite_semigroups.py"
sh: kpsewhich: not found
sh: kpsewhich: not found
**********************************************************************
File "/export/home/drkirkby/32/sage-4.3.4.alpha0/devel/sage/sage/categories/finite_semigroups.py", line 232:
    sage: sorted(S.j_transversal_of_idempotents())
Expected:
    ['a', 'ab', 'ac', 'acb', 'b', 'bc', 'c']
Got:
    ['a', 'ab', 'ac', 'acb', 'b', 'c', 'cb']

So there are two problems.

• kpsewhich: not found
• doctest failure

But #8180 was supposed to fix this kpsewhich issue, so I believe the fix is not working fully.

I'll create a ticket for the test failure if needed. But I believe I see this mentioned on sage-devel, so it looks like I'm not the only one with this issue. The relevant ticket reporting this issue is #8430.

CC: @sagetrac-mvngu @jaapspies @jhpalmieri @qed777 @sagetrac-sage-combinat

Component: porting: Solaris

Author: John Palmieri

Reviewer: Mike Hansen, David Kirkby

Merged: sage-4.3.4.alpha1

Issue created by migration from https://trac.sagemath.org/ticket/8445

[sagetrac-mvngu]

comment:1

The sage-combinat team might be interested in this ticket.

[jhpalmieri]

Author: John Palmieri

[jhpalmieri]

comment:3

Here's a patch. It's hard for me to doctest it on Solaris: I think the only machine I have access to is t2.math, but there isn't enough room in /scratch for me to install Sage. So please test it out. It works for me on several other machines, and on t2, if I load a file containing the relevant code, that works as well.

The patch also reformats the warning messages that get printed if tkz-berge.sty (etc.) are not present.

[sagetrac-drkirkby]

comment:4

Attachment: trac_8445-kpsewhich-solaris.patch.gz

Thank's John,

I will test this, but it might take me a few days, as I have a very busy schedule this week.

Fortunately this is not a critical patch.

Dave

[sagetrac-drkirkby]

comment:5

John,

I've tested this on Solaris, and find no more "kpsewhich" problems. So from my point of view it is working.

However, I don't feel comfortable giving this a positive review, as I don't understand much of the code.

Perhaps one of the others cc'ed on the ticket can look over this, keeping in mind that it does solve the problem I reported.

Dave

[mwhansen]

comment:6

This change looks good to me.

[mwhansen]

Reviewer: Mike Hansen, David Kirkby

[mwhansen]

Merged: sage-4.3.4.alpha1