Upgrade to pari 2.11.4 #29313

timokau · 2020-03-11T10:38:29Z

Unfortunately the update is not trivial. Here are some doctest failures:

File "/nix/store/vqiiwb9rzmzh97gzx1wmvqyiy9d5qhxj-sage-src-8.9/src/sage/rings/number_field/number_field.py", line 3253, in sage.rings.number_field.number_field.NumberField_generic.ideals_of_bdd_norm
Failed example:
    for n in d:
        print(n)
        for I in d[n]:
            print(I)
Expected:
    1
    Fractional ideal (1)
    2
    Fractional ideal (2, 1/2*a - 1/2)
    Fractional ideal (2, 1/2*a + 1/2)
    3
    Fractional ideal (3, 1/2*a - 1/2)
    Fractional ideal (3, 1/2*a + 1/2)
    4
    Fractional ideal (4, 1/2*a + 3/2)
    Fractional ideal (2)
    Fractional ideal (4, 1/2*a + 5/2)
    5
    6
    Fractional ideal (1/2*a - 1/2)
    Fractional ideal (6, 1/2*a + 5/2)
    Fractional ideal (6, 1/2*a + 7/2)
    Fractional ideal (1/2*a + 1/2)
    7
    8
    Fractional ideal (1/2*a + 3/2)
    Fractional ideal (4, a - 1)
    Fractional ideal (4, a + 1)
    Fractional ideal (1/2*a - 3/2)
    9
    Fractional ideal (9, 1/2*a + 11/2)
    Fractional ideal (3)
    Fractional ideal (9, 1/2*a + 7/2)
    10
Got:
    1
    Fractional ideal (1)
    2
    Fractional ideal (2, 1/2*a - 1/2)
    Fractional ideal (2, 1/2*a + 1/2)
    3
    Fractional ideal (3, 1/2*a + 1/2)
    Fractional ideal (3, 1/2*a - 1/2)
    4
    Fractional ideal (4, 1/2*a + 3/2)
    Fractional ideal (2)
    Fractional ideal (4, 1/2*a + 5/2)
    5
    6
    Fractional ideal (6, 1/2*a + 7/2)
    Fractional ideal (1/2*a + 1/2)
    Fractional ideal (1/2*a - 1/2)
    Fractional ideal (6, 1/2*a + 5/2)
    7
    8
    Fractional ideal (1/2*a + 3/2)
    Fractional ideal (4, a - 1)
    Fractional ideal (4, a + 1)
    Fractional ideal (1/2*a - 3/2)
    9
    Fractional ideal (9, 1/2*a + 7/2)
    Fractional ideal (3)
    Fractional ideal (9, 1/2*a + 11/2)
    10
**********************************************************************
File "/nix/store/vqiiwb9rzmzh97gzx1wmvqyiy9d5qhxj-sage-src-8.9/src/sage/rings/number_field/number_field.py", line 4369, in sage.rings.number_field.number_field.NumberField_generic._S_class_group_and_units
Failed example:
    K._S_class_group_and_units(tuple(K.primes_above(13)))
Expected:
    ([2/13*a^2 + 1/13*a - 677/13,
      1/13*a^2 + 7/13*a - 332/13,
      -1/13*a^2 + 6/13*a + 345/13,
      -1,
      2/13*a^2 + 1/13*a - 755/13,
      1/13*a^2 - 19/13*a - 7/13],
     [(Fractional ideal (11, a - 2), 2), (Fractional ideal (19, a + 7), 2)])
Got:
    ([2/13*a^2 + 1/13*a - 677/13,
      1/13*a^2 + 7/13*a - 332/13,
      -1/13*a^2 + 6/13*a + 345/13,
      -1,
      2/13*a^2 + 1/13*a - 755/13,
      1/13*a^2 + 20/13*a - 7/13],
     [(Fractional ideal (11, a - 2), 2), (Fractional ideal (19, a + 7), 2)])
**********************************************************************
File "/nix/store/vqiiwb9rzmzh97gzx1wmvqyiy9d5qhxj-sage-src-8.9/src/sage/rings/number_field/number_field.py", line 4527, in sage.rings.number_field.number_field.NumberField_generic.selmer_group
Failed example:
    K.selmer_group(K.primes_above(13), 2)
Expected:
    [2/13*a^2 + 1/13*a - 677/13,
     1/13*a^2 + 7/13*a - 332/13,
     -1/13*a^2 + 6/13*a + 345/13,
     -1,
     2/13*a^2 + 1/13*a - 755/13,
     1/13*a^2 - 19/13*a - 7/13,
     -1/13*a^2 + 45/13*a - 97/13,
     2/13*a^2 + 40/13*a - 27/13]
Got:
    [2/13*a^2 + 1/13*a - 677/13,
     1/13*a^2 + 7/13*a - 332/13,
     -1/13*a^2 + 6/13*a + 345/13,
     -1,
     2/13*a^2 + 1/13*a - 755/13,
     1/13*a^2 + 20/13*a - 7/13,
     1/13*a^2 - 45/13*a + 97/13,
     -2/13*a^2 - 40/13*a + 27/13]
**********************************************************************
File "/nix/store/vqiiwb9rzmzh97gzx1wmvqyiy9d5qhxj-sage-src-8.9/src/sage/rings/number_field/number_field.py", line 5179, in sage.rings.number_field.number_field.NumberField_generic.elements_of_norm
Failed example:
    K.elements_of_norm(7)
Expected:
    [7/225*a^2 - 7/75*a - 42/25]
Got:
    [28/225*a^2 + 77/75*a - 133/25]
**********************************************************************
File "/nix/store/vqiiwb9rzmzh97gzx1wmvqyiy9d5qhxj-sage-src-8.9/src/sage/rings/number_field/number_field.py", line 5282, in sage.rings.number_field.number_field.NumberField_generic.factor
Failed example:
    f = L.factor(a + 1); f
Expected:
    (Fractional ideal (1/2*a*b - a + 1/2)) * (Fractional ideal (-1/2*a*b - a + 1/2))
Got:
    (Fractional ideal (-1/2*b + 1/2*a + 1)) * (Fractional ideal (-1/2*a*b - a + 1/2))
**********************************************************************
File "/nix/store/vqiiwb9rzmzh97gzx1wmvqyiy9d5qhxj-sage-src-8.9/src/sage/rings/number_field/number_field.py", line 6359, in sage.rings.number_field.number_field.NumberField_generic.uniformizer
Failed example:
    [K.uniformizer(P) for P,e in factor(K.ideal(7))]
Expected:
    [t^2 + 3*t + 1]
Got:
    [t^2 - 4*t + 1]
**********************************************************************
File "/nix/store/vqiiwb9rzmzh97gzx1wmvqyiy9d5qhxj-sage-src-8.9/src/sage/rings/number_field/number_field.py", line 6615, in sage.rings.number_field.number_field.NumberField_generic.S_unit_group
Failed example:
    U.log(u)
Expected:
    (1, 1, 4, 1, 5)
Got:
    (0, 1, 4, 1, 5)
**********************************************************************
File "/nix/store/vqiiwb9rzmzh97gzx1wmvqyiy9d5qhxj-sage-src-8.9/src/sage/rings/number_field/number_field.py", line 7599, in sage.rings.number_field.number_field.NumberField_absolute.optimized_representation
Failed example:
    K.optimized_representation()
Expected:
    (Number Field in a1 with defining polynomial x^3 - 7*x - 7,
     Ring morphism:
       From: Number Field in a1 with defining polynomial x^3 - 7*x - 7
       To:   Number Field in a with defining polynomial 7/9*x^3 + 7/3*x^2 - 56*x + 123
       Defn: a1 |--> 7/225*a^2 - 7/75*a - 42/25,
     Ring morphism:
       From: Number Field in a with defining polynomial 7/9*x^3 + 7/3*x^2 - 56*x + 123
       To:   Number Field in a1 with defining polynomial x^3 - 7*x - 7
       Defn: a |--> -15/7*a1^2 + 9)
Got:
    (Number Field in a1 with defining polynomial x^3 - 7*x - 7, Ring morphism:
       From: Number Field in a1 with defining polynomial x^3 - 7*x - 7
       To:   Number Field in a with defining polynomial 7/9*x^3 + 7/3*x^2 - 56*x + 123
       Defn: a1 |--> 28/225*a^2 + 77/75*a - 133/25, Ring morphism:
       From: Number Field in a with defining polynomial 7/9*x^3 + 7/3*x^2 - 56*x + 123
       To:   Number Field in a1 with defining polynomial x^3 - 7*x - 7
       Defn: a |--> -60/7*a1^2 + 15*a1 + 39)
**********************************************************************
8 items had failures:
   1 of  10 in sage.rings.number_field.number_field.NumberField_absolute.optimized_representation
   1 of  27 in sage.rings.number_field.number_field.NumberField_generic.S_unit_group
   1 of   9 in sage.rings.number_field.number_field.NumberField_generic._S_class_group_and_units
   1 of   6 in sage.rings.number_field.number_field.NumberField_generic.elements_of_norm
   1 of  15 in sage.rings.number_field.number_field.NumberField_generic.factor
   1 of   4 in sage.rings.number_field.number_field.NumberField_generic.ideals_of_bdd_norm
   1 of  32 in sage.rings.number_field.number_field.NumberField_generic.selmer_group
   1 of  20 in sage.rings.number_field.number_field.NumberField_generic.uniformizer
    [2076 tests, 8 failures, 52.52 s]
----------------------------------------------------------------------
sage -t --long /nix/store/vqiiwb9rzmzh97gzx1wmvqyiy9d5qhxj-sage-src-8.9/src/sage/rings/number_field/number_field.py  # 8 doctests failed

I will not do the update myself in this ticket, I only opened it to notify people of the issues.

Tarfile: see checksums_ini [upstream_url]

Depends on #29472

CC: @antonio-rojas @kiwifb @saraedum @tobihan @fchapoton @embray @jdemeyer @dimpase

Component: packages: standard

Author: Antonio Rojas

Branch/Commit: 549e094

Reviewer: Dima Pasechnik, François Bissey

Issue created by migration from https://trac.sagemath.org/ticket/29313

The text was updated successfully, but these errors were encountered:

antonio-rojas · 2020-03-11T10:51:24Z

comment:1

Already updated on Arch. Some of the failures are actual incorrect results which are due to a pari bug fixed in [1]. The remaining ones are due to changes in chosen ideal generators, and in computation precision. Those just need the doctests to be updated.

[1] https://pari.math.u-bordeaux.fr/cgi-bin/gitweb.cgi?p=pari.git;a=commitdiff;h=c7a1d35f382e96ddf14694be27a0ca5746880700

Will post a branch when I have some free time unless someone beats me to it.

timokau · 2020-03-11T10:53:09Z

comment:2

I just now noticed #28424 (missed it initially because I searched for pari 2.11.3 in particular). That ticket proposes an update to 2.12 though, which is still in beta and probably harder to do. So I think we can keep this open.

timokau · 2020-03-11T10:56:40Z

comment:3

Replying to @antonio-rojas:

Already updated on Arch. Some of the failures are actual incorrect results which are due to a pari bug fixed in [1]. The remaining ones are due to changes in chosen ideal generators, and in computation precision. Those just need the doctests to be updated.

[1] https://pari.math.u-bordeaux.fr/cgi-bin/gitweb.cgi?p=pari.git;a=commitdiff;h=c7a1d35f382e96ddf14694be27a0ca5746880700

Will post a branch when I have some free time unless someone beats me to it.

Thats great news, amazing how fast you always come up with those fixes :) For anyone else wondering, here is Antonio's patch.

antonio-rojas · 2020-03-11T10:57:13Z

comment:4

Since Sage can now use system pari, perhaps 2.11.3 should be blacklisted until this is merged. Ideally, distros shipping 2.11.3 should backport the patch as it is quite a serious bug.

timokau · 2020-03-11T11:23:24Z

comment:5

Oh, I misread your original comment and I thought that you meant the tests were wrong before and 2.11.3 fixes the bug. I'll backport the pari fix for nix.

dimpase · 2020-03-11T11:39:41Z

comment:6

Pari 2.11.3 has been released just few days ago, and that bug was fixed last year. Could it be that upstream forgot to include the fix?

What's an example of incorrect maths output without that patch?

antonio-rojas · 2020-03-11T11:43:23Z

comment:7

Replying to @dimpase:

Pari 2.11.3 has been released just few days ago, and that bug was fixed last year. Could it be that upstream forgot to include the fix?

Yes, they forgot to backport the fix to stable. See [1]

What's an example of incorrect maths output without that patch?

You have some examples in [1] and [2]

[1] https://pari.math.u-bordeaux.fr/cgi-bin/bugreport.cgi?bug=2208

[2] https://pari.math.u-bordeaux.fr/cgi-bin/bugreport.cgi?bug=2164

This causes a few failures in Sage. Among other things, simon's scripts are totally broken.

timokau · 2020-03-11T17:36:40Z

comment:8

When patching pari with [1] and sage with [2] I still get at least one doctest failure (I only tested the number_fields.py file for now):

sage -t --long /nix/store/nsvlj78rfc0s8077drna9xw54xw2z50v-sage-src-8.9/src/sage/rings/number_field/number_field.py
**********************************************************************
File "/nix/store/nsvlj78rfc0s8077drna9xw54xw2z50v-sage-src-8.9/src/sage/rings/number_field/number_field.py", line 6615, in sage.rings.number_field.number_field.NumberField_generic.S_unit_group
Failed example:
    U.log(u)
Expected:
    (1, 1, 4, 1, 5)
Got:
    (0, 1, 4, 1, 5)
**********************************************************************
1 item had failures:
   1 of  27 in sage.rings.number_field.number_field.NumberField_generic.S_unit_group
    [2076 tests, 1 failure, 54.23 s]

Am I missing something?

[1] https://pari.math.u-bordeaux.fr/cgi-bin/gitweb.cgi?p=pari.git;a=patch;h=c7a1d35f382e96ddf14694be27a0ca5746880700

[2] https://aur.archlinux.org/cgit/aur.git/plain/sagemath-pari-2.11.3.patch?h=sagemath-git&id=02e1d58bd1cd70935d69a4990469d18be6bd2c43

kiwifb · 2020-03-11T19:58:46Z

comment:9

Given that this seem to be a serious bug, can we expect upstream to release 2.11.4 shortly to fix it?

antonio-rojas · 2020-03-12T08:24:15Z

comment:10

Replying to @timokau:

When patching pari with [1] and sage with [2] I still get at least one doctest failure (I only tested the number_fields.py file for now):

sage -t --long /nix/store/nsvlj78rfc0s8077drna9xw54xw2z50v-sage-src-8.9/src/sage/rings/number_field/number_field.py
**********************************************************************
File "/nix/store/nsvlj78rfc0s8077drna9xw54xw2z50v-sage-src-8.9/src/sage/rings/number_field/number_field.py", line 6615, in sage.rings.number_field.number_field.NumberField_generic.S_unit_group
Failed example:
    U.log(u)
Expected:
    (1, 1, 4, 1, 5)
Got:
    (0, 1, 4, 1, 5)
**********************************************************************
1 item had failures:
   1 of  27 in sage.rings.number_field.number_field.NumberField_generic.S_unit_group
    [2076 tests, 1 failure, 54.23 s]

Am I missing something?

That's one of the failures that should be fixed with pari's patch, are you sure you're running the patched version?

timokau · 2020-03-12T10:26:21Z

comment:11

No, I wasn't. I had a typo ("patch" instead of "patches") that resulted in the patch silently being ignored. Sorry for the noise.

As an aside, the upstream patch doesn't quite apply cleanly but the arch patch [1] is rebased. Thanks!

[1] https://git.archlinux.org/svntogit/community.git/plain/repos/community-x86_64/c7a1d35f.patch?h=packages/pari&id=27893d227290dc3821d68aa25877d9765c204dad

antonio-rojas · 2020-03-13T23:37:23Z

Branch: u/arojas/upgrade_to_pari_2_11_3

antonio-rojas · 2020-03-13T23:41:08Z

Commit: ce42f11

antonio-rojas · 2020-03-13T23:41:08Z

comment:13

Needs testing for the check_functional_equation() output on 32 bits, i've only checked 64 bits. Also, some elliptic curves expert should check if the changes in elliptic_curves/ell_number_field.py are mathematically correct. All other ones seem OK to me.

New commits:

`4ba1ea8`	`Update pari to 2.11.3`
`d934a82`	`Remove wrong patch`
`ce42f11`	`Adjust doctests for new output with pari 2.11.3`

antonio-rojas · 2020-03-13T23:41:08Z

Author: Antonio Rojas

kliem · 2020-04-03T18:18:04Z

comment:14

The way I understand the configuration, once this ticket is through, it will reject a system pari unless it is at least version 2.11.3. Is this correct?

antonio-rojas · 2020-04-03T20:02:17Z

comment:15

Replying to @kliem:

The way I understand the configuration, once this ticket is through, it will reject a system pari unless it is at least version 2.11.3. Is this correct?

No, the minimum pari system version is defined in spkg-configure and is independent of the version bundled in sage

kliem · 2020-04-03T20:10:05Z

comment:16

Ok, then I get why you want sage's test to be work with pari 2.11.3 and previous versions.

Do you want this to be done in this ticket or in a seperate ticket?

antonio-rojas · 2020-04-04T12:00:40Z

comment:17

Replying to @kliem:

Ok, then I get why you want sage's test to be work with pari 2.11.3 and previous versions.

Do you want this to be done in this ticket or in a seperate ticket?

I don't really mind as long as I can get rid of my downstream patch. Feel free to take over this branch and modify the test results so they pass with both versions.

kliem · 2020-04-06T16:25:40Z

Dependencies: #29472

sagetrac-git · 2020-04-18T12:12:18Z

Changed commit from ce42f11 to 507576f

sagetrac-git · 2020-04-18T12:12:18Z

Branch pushed to git repo; I updated commit sha1. New commits:

`507576f`	`Update to 2.11.4`

sagetrac-git · 2020-06-19T11:58:17Z

Changed commit from 507576f to 4a23a12

sagetrac-git · 2020-06-19T11:58:17Z

Branch pushed to git repo; I updated commit sha1. New commits:

`4a23a12`	`Merge branch 'develop' of git://git.sagemath.org/sage into t/29313/upgrade_to_pari_2_11_3`

sagetrac-git · 2020-06-19T12:00:10Z

Changed commit from 4a23a12 to dfcdcb0

sagetrac-git · 2020-06-19T12:00:10Z

Branch pushed to git repo; I updated commit sha1. New commits:

`dfcdcb0`	`Drop test fixes, handled in #29472`

dimpase · 2020-06-19T13:07:11Z

comment:25

please also apply

--- a/build/pkgs/pari/checksums.ini
+++ b/build/pkgs/pari/checksums.ini
@@ -2,3 +2,4 @@ tarball=pari-VERSION.tar.gz
 sha1=2b9ff51feb388664b834dc346a44867546c78618
 md5=fb2968d7805424518fe44a59a2024afd
 cksum=1247903778
+upstream_url=https://pari.math.u-bordeaux.fr/pub/pari/unix/pari-2.11.4.tar.gz

sagetrac-git · 2020-06-19T13:16:37Z

Branch pushed to git repo; I updated commit sha1. New commits:

`549e094`	`Add upstream url`

sagetrac-git · 2020-06-19T13:16:37Z

Changed commit from dfcdcb0 to 549e094

dimpase · 2020-06-19T19:06:52Z

comment:28

some tests on macOS and Cygwin would be great. Perhaps adapting GH Actions to restrict platforms...

mkoeppe · 2020-06-19T19:55:33Z

comment:29

Replying to @dimpase:

Perhaps adapting GH Actions to restrict platforms...

There's a ticket for that (#29535 - Customize CI workflows on GitHub Actions using specially structured tag names)

dimpase · 2020-06-19T21:38:51Z

comment:30

for some reason I don't get macOS runs on GH Actions any more. Is it by design, or GH pulled them?

mkoeppe · 2020-06-19T22:52:47Z

comment:31

Replying to @dimpase:

for some reason I don't get macOS runs on GH Actions any more. Is it by design, or GH pulled them?

No, works just fine for me -- for example here: https://github.com/mkoeppe/sage/actions/runs/141135018

mkoeppe · 2020-06-19T22:54:37Z

comment:32

You probably need to cancel some of your old queued workflows such as this one: https://github.com/dimpase/sage/runs/784000040?check_suite_focus=true

mkoeppe · 2020-07-19T05:44:44Z

comment:33

What's missing on this ticket?

dimpase · 2020-07-19T07:51:04Z

comment:34

basically, someone should bite the bullet and make pari-related doctests less rigid (cf. e.g. the 1st example in the ticket description - it assumes an output order which is not really fixed),so that different versions of pari would still pass.

antonio-rojas · 2020-07-19T08:32:02Z

comment:35

Replying to @dimpase:

basically, someone should bite the bullet and make pari-related doctests less rigid (cf. e.g. the 1st example in the ticket description - it assumes an output order which is not really fixed),so that different versions of pari would still pass.

That's done in #29472 already

dimpase · 2020-07-19T08:35:27Z

comment:36

ah, ok, then in principle it should be good to go.

kiwifb · 2020-07-19T08:49:28Z

comment:37

Been using it for a little while in sage-on-gentoo and I don't need any patches for doctests anymore for some time now. So, I am putting this to positive.

kiwifb · 2020-07-19T08:49:28Z

Changed reviewer from Dima Pasechnik to Dima Pasechnik, François Bissey

vbraun · 2020-07-28T22:34:31Z

Changed branch from u/arojas/upgrade_to_pari_2_11_3 to 549e094

timokau added this to the sage-9.1 milestone Mar 11, 2020

timokau added c: packages: standard labels Mar 11, 2020

antonio-rojas added the s: needs review label Mar 13, 2020

antonio-rojas changed the title ~~Upgrade to pari 2.11.3~~ Upgrade to pari 2.11.4 Apr 18, 2020

mkoeppe modified the milestones: sage-9.1, sage-9.2 May 4, 2020

dimpase added s: needs work and removed s: needs review labels Jun 19, 2020

This comment has been minimized.

Sign in to view

dimpase added s: needs review and removed s: needs work labels Jun 19, 2020

kiwifb added s: positive review and removed s: needs review labels Jul 19, 2020

vbraun removed the s: positive review label Jul 28, 2020

vbraun closed this as completed in d01edd9 Jul 28, 2020

kliem mentioned this issue Jun 22, 2020

Modify doctest such that they work with pari >= 2.11.3 and pari < 2.11.3 #29472

Closed

slel mentioned this issue Jul 2, 2021

Upgrade: pari 2.13 #30801

Closed

Upgrade to pari 2.11.4 #29313

Upgrade to pari 2.11.4 #29313

Comments

timokau commented Mar 11, 2020

antonio-rojas commented Mar 11, 2020

timokau commented Mar 11, 2020

timokau commented Mar 11, 2020

antonio-rojas commented Mar 11, 2020

timokau commented Mar 11, 2020

dimpase commented Mar 11, 2020

antonio-rojas commented Mar 11, 2020

timokau commented Mar 11, 2020

kiwifb commented Mar 11, 2020

antonio-rojas commented Mar 12, 2020

timokau commented Mar 12, 2020

antonio-rojas commented Mar 13, 2020

antonio-rojas commented Mar 13, 2020

antonio-rojas commented Mar 13, 2020

antonio-rojas commented Mar 13, 2020

kliem commented Apr 3, 2020

antonio-rojas commented Apr 3, 2020

kliem commented Apr 3, 2020

antonio-rojas commented Apr 4, 2020

kliem commented Apr 6, 2020

sagetrac-git mannequin commented Apr 18, 2020

sagetrac-git mannequin commented Apr 18, 2020

sagetrac-git mannequin commented Jun 19, 2020

sagetrac-git mannequin commented Jun 19, 2020

sagetrac-git mannequin commented Jun 19, 2020

sagetrac-git mannequin commented Jun 19, 2020

dimpase commented Jun 19, 2020

sagetrac-git mannequin commented Jun 19, 2020

sagetrac-git mannequin commented Jun 19, 2020

This comment has been minimized.

dimpase commented Jun 19, 2020

mkoeppe commented Jun 19, 2020

dimpase commented Jun 19, 2020

mkoeppe commented Jun 19, 2020

mkoeppe commented Jun 19, 2020

mkoeppe commented Jul 19, 2020

dimpase commented Jul 19, 2020

antonio-rojas commented Jul 19, 2020

dimpase commented Jul 19, 2020

kiwifb commented Jul 19, 2020

kiwifb commented Jul 19, 2020

vbraun commented Jul 28, 2020