Opened 9 years ago

Closed 9 years ago

Last modified 9 years ago

#8453 closed defect (fixed)

Update PARI to 2.3.5

Reported by: mhansen Owned by: tbd
Priority: major Milestone: sage-4.4.1
Component: packages: standard Keywords:
Cc: cremona Merged in: sage-4.4.1.alpha0
Authors: Mike Hansen Reviewers: John Cremona, David Loeffler
Report Upstream: N/A Work issues:
Branch: Commit:
Dependencies: Stopgaps:

Attachments (2)

trac_8453.patch (2.3 KB) - added by mhansen 9 years ago.
trac_8453-reviewer.patch (3.6 KB) - added by davidloeffler 9 years ago.
apply over previous patch

Download all attachments as: .zip

Change History (12)

Changed 9 years ago by mhansen

comment:1 Changed 9 years ago by mhansen

  • Status changed from new to needs_review

comment:2 Changed 9 years ago by cremona

Mike, I will give this a try. Unfortunately I have not been able to build 4.3.3 on my 64-bit machine (it builds but thousands of tests fail) so I'll be limited to testing on a 32-bit.

Can we try to collect from other tickets problems which have been attributed to bugs in the pari library, to see if this upgrade fixes them? If so, there ought tobe appropriate doctests to prove it, and cross-referencing the other tickets, which might therefore be closed.

comment:3 Changed 9 years ago by cremona

  • Status changed from needs_review to needs_work

Successfully installed the new spkg and patch on 4.3.4.alpha0 on a 32-bit ubuntu machine. Tested all library (-long) and found a few failures which should be easy to fix (especially the last one!):

sage -t -long sage/functions/transcendental.py
**********************************************************************
File "/home/john/sage-4.3.4.alpha0/devel/sage-pari/sage/functions/transcendental.py", line 78:
    sage: w = exponential_integral_1(2,4); w
Expected:
    [0.048900510708061118, 0.003779352409848905, 0.00036008245216265542, 3.7665622843921715e-05] 
Got:
    [0.048900510708061118, 0.0037793524098489067, 0.00036008245216265873, 3.7665622843924751e-05]
**********************************************************************
File "/home/john/sage-4.3.4.alpha0/devel/sage-pari/sage/functions/special.py", line 1456:
    sage: exp_int(6)
Expected:
    doctest:...: DeprecationWarning: The method expint() is deprecated. Use -Ei(-x) or exponential_integral_1(x) as needed instead.
    0.000360082452162655
Got:
    doctest:1: DeprecationWarning: The method expint() is deprecated. Use -Ei(-x) or exponential_integral_1(x) as needed instead.
    0.000360082452162659
**********************************************************************
File "/home/john/sage-4.3.4.alpha0/devel/sage-pari/sage/libs/pari/gen.pyx", line 7848:
    sage: E.ellwp(1, flag=2)
Expected:
    [14.2992028590818 + 1.140149682 E-18*I, 50.0619300880073 + 1.040834085 E-17*I] 
Got:
    [14.2992028590818 + 0.E-18*I, 50.0619300880073 - 3.469446952 E-18*I]
**********************************************************************
File "/home/john/sage-4.3.4.alpha0/devel/sage-pari/sage/interfaces/gp.py", line 476:
    sage: gp.version()
Expected:
    ((2, 3, 3), 'GP/PARI CALCULATOR Version 2.3.3 (released)')
Got:
    ((2, 3, 5), 'GP/PARI CALCULATOR Version 2.3.5 (released)')

Changed 9 years ago by davidloeffler

apply over previous patch

comment:4 follow-up: Changed 9 years ago by davidloeffler

  • Status changed from needs_work to needs_review

I have added a second patch which changes the doctests mentioned above. They should now pass on both 64-bit and 32-bit (the latter using the output John got above). I've also put in a doctest to confirm that #8415 is fixed.

John, can you double-check that it now works on 32-bit? I've marked it as "needs review" for now, but if that passes, I think we can give it a positive review (and mark #8415 as fixed as well). Sadly, #7736 is still broken.

David

comment:5 in reply to: ↑ 4 Changed 9 years ago by cremona

Replying to davidloeffler:

I have added a second patch which changes the doctests mentioned above. They should now pass on both 64-bit and 32-bit (the latter using the output John got above). I've also put in a doctest to confirm that #8415 is fixed.

Excellent

John, can you double-check that it now works on 32-bit? I've marked it as "needs review" for now, but if that passes, I think we can give it a positive review (and mark #8415 as fixed as well). Sadly, #7736 is still broken.

OK, testing now....

David

comment:6 Changed 9 years ago by cremona

  • Reviewers set to John Cremona, David Loeffler
  • Status changed from needs_review to positive_review

All pass on 32-bit.

comment:7 Changed 9 years ago by davidloeffler

  • Milestone set to sage-4.3.4

FWIW: I've just checked and it *does* build and run OK on Solaris, by the way. I haven't done a full doctest run because that would take a looong time; but I tested a selection of relevant files including sage/rings/number_field, sage/functions/transcendental and sage/libs/pari/gen and they all seem to pass.

comment:8 Changed 9 years ago by was

  • Merged in set to 4.4.1.alpha0
  • Resolution set to fixed
  • Status changed from positive_review to closed

comment:9 Changed 9 years ago by mvngu

Since Pari is now upgraded to 2.3.5, we no longer need ticket #7979 for patching Pari 2.3.3. That ticket concerns the case where Pari 2.3.3 sometimes ignores the build option "--graphic=none". However, the file config/get_fltk in Pari 2.3.5 has the same logic as in Pari 2.3.3 so it's possible that Pari 2.3.5 also ignores the build option "--graphic=none". If that issue comes up, open another ticket to patch Pari 2.3.5 for Sage.

comment:10 Changed 9 years ago by mvngu

  • Merged in changed from 4.4.1.alpha0 to sage-4.4.1.alpha0
Note: See TracTickets for help on using tickets.