Doctest failure in SageNB's sageinspect.py with #8988

[mpatel]

This happens with the patches at #8988, which I've merged in the forthcoming 4.5.2.alpha0:

$ ./sage -t -long  local/lib/python2.6/site-packages/sagenb-0.8.1-py2.6.egg/sagenb/misc/sageinspect.py
sage -t -long "local/lib/python2.6/site-packages/sagenb-0.8.1-py2.6.egg/sagenb/misc/sageinspect.py"
**********************************************************************
File "/mnt/usb1/scratch/mpatel/tmp/sage-4.5.1-rm/local/lib/python2.6/site-packages/sagenb-0.8.1-py2.6.egg/sagenb/misc/sageinspect.py", line 997:
    sage: sage_getvariablename(A)
Expected:
    ['A', 'B']
Got:
    ['B', 'A']
**********************************************************************
1 items had failures:
   1 of   8 in __main__.example_22
***Test Failed*** 1 failures.

[vbraun]

Can you double-check that you applied trac_8988-sageinspect_doctest_fix.patch from #8988? That patch was supposed to fix precisely this issue. In fact, there are 3 patches in #8988, see ​http://trac.sagemath.org/sage_trac/ticket/8988#comment:42

[mpatel]

I've applied the 3 patches mentioned in #8988 comment 42. The failure above is in the notebook's sageinspect.py file, which is very similar to the Sage library's version.

Of course, we should try to reconcile the two files, to the extent it's permitted by SageNB's status as an independent project. I believe we've been updating them together when one changes, in order to keep both current.

[mpatel]

I can make the SageNB patch a bit later this week and try to ensure it's merged into 4.5.2.alphaX.

[leif]

SageNB repo patch - logically same as ​http://trac.sagemath.org/sage_trac/attachment/ticket/8988/trac_8988-sageinspect_doctest_fix.patch

[leif]

Please excuse if my upload violates any rules regarding Sage-SageNB collaboration...

The patch is almost the same as the one to the Sage library here: ​http://trac.sagemath.org/sage_trac/attachment/ticket/8988/trac_8988-sageinspect_doctest_fix.patch

I've built a SageNB 0.8.1.p1 spkg, installed it on 4.5.2.alpha0, and ./sage -t -long -sagenb passed all tests.

(I'm not going to upload the patched spkg though, it's almost 19MB. So if appropriate, someone else has to provide it on sage.math.)

-Leif

[leif]

P.S.:

leif@portland:~/Sage/spkgs/sagenb-0.8.1.p1/src/sagenb$ hg status
? sagenb/sagenb.po

(This is of course in vanilla 0.8.1, too.)

[ddrake]

Replying to vbraun:

Can you double-check that you applied trac_8988-sageinspect_doctest_fix.patch from #8988? That patch was supposed to fix precisely this issue. In fact, there are 3 patches in #8988, see ​http://trac.sagemath.org/sage_trac/ticket/8988#comment:42

All three patches were merged in 4.5.2.alpha0 -- changesets 14652-4.

[vbraun]

Yes, as mpatel already pointed out, the problematic doctest is in sagenb/misc/sageinspect.py and not in sage/misc/sageinspect.py (sagenb vs. sage). The latter is fixed by the patch in #8988, and the former is fixed by leif's trac_9554-fix_order_of_var_names-SageNB-repo.patch. Both patches make identical changes to code that is duplicated in the notebook spkg.

Since the patch is clearly the right thing to do I'll give it a positive review. Whoever does the next SageNB release, please add this patch. The tracker bug for SageNB 0.8.2 is #9572 where this patch is already listed.

[mpatel]

Replying to leif:

P.S.:

leif@portland:~/Sage/spkgs/sagenb-0.8.1.p1/src/sagenb$ hg status
? sagenb/sagenb.po

(This is of course in vanilla 0.8.1, too.)

This is now #9580.

[mpatel]

I've merged the patch into SageNB 0.8.2, which awaits review at #9572.

[mpatel]

Replying to leif:

(I'm not going to upload the patched spkg though, it's almost 19MB. So if appropriate, someone else has to provide it on sage.math.)

Do you have a Sage cluster (i.e., sage.math, boxen.math, etc.) account? If not, William Stein can make one for you.