I can further add that GAP 4.9 fixes this: for
computing NormalSubgroups(SmallGroup(1458,1180));
CRISP is no longer neeed.
On Saturday, February 17, 2018 at 10:30:01 AM UTC, Dima Pasechnik wrote:
>
>
>
> On Saturday, February 17, 2018 at 7:31:19 AM UTC, David Roe wrote:
>>
>> I realized that I ran
>> sage: gap.eval('LoadAllPackages();')
>> which broke libgap.
>>
>
> Really? I don't think so. It didn't break it for me, although it did break
> GAP's workspaces, which is for some reason very fragile.
> (libGAP and GAP have separate workspaces in Sage)
> What happened is the following, I think: copying the packages from GAP
> source broke some GAP packages in Sage's
> GAP pkg/ directory, namely at least sonata (mea culpa, I should have
> warned you about it)
>
> sage: gap_reset_workspace(verbose=True)
> Loading GAP package GAPDoc
> Loading GAP package ctbllib
> Loading GAP package sonata
> Warning: this should never happen
>
> (and Sage hangs)
>
> In fact, this is coming from
>
> elif x==13: #@w GAP is trying to send a Window command
>
> in src/sage/interfaces/gap.py
> I have no clue what "Window command" is. Anyway, the error message should
> not be identical to the one from
>
> elif x==12: #@sN shouldn't happen
>
> I also don't know the reason for that "Window command" --- is it cause it
> is meant to print some info message?
>
>
>
>> For anyone else reading this, the following fixed the problem (taken from
>> gap_reset_workspace, with sonata and guava removed due to "Warning: this
>> should never happen" and braid removed because I don't have it installed):
>>
>> g = Gap(use_workspace_cache=False, max_workspace_size=None)
>> g.eval('SetUserPreference("HistoryMaxLines", 30)')
>> for pkg in ['GAPDoc', 'ctbllib', 'factint', 'gapdoc', 'grape', 'design',
>> 'toric', 'laguna']:
>> g.load_package(pkg, verbose=True)
>> g.save_workspace()
>> g.quit()
>>
>> Thanks for the help Dima: my calculations are running smoothly again!
>> David
>>
>> On Sat, Feb 17, 2018 at 12:57 AM, David Roe <roed...@gmail.com> wrote:
>>
>>> Thanks, that helps a lot. After some experimentation, it seems like
>>> gap> LoadPackage("CRISP");
>>> is sufficient.
>>>
>>> However, when I try to do it from sage I get the following:
>>> sage: gap.eval('LoadPackage("CRISP");')
>>> Warning: this should never happen
>>>
>>> and sage hangs. Googling leads me to
>>> https://groups.google.com/forum/#!topic/sage-support/60HHHL8Bw_8, which
>>> I interpret as suggesting that CRISP is using LoadDynamicModule. However,
>>> searching on https://github.com/bh11/crisp, I don't see any calls to
>>> that function. Is there something else that could be responsible? Is
>>> there a way to use CRISP through gap.eval in Sage?
>>> David
>>>
>>> On Fri, Feb 16, 2018 at 11:40 PM, Dima Pasechnik <dim...@gmail.com>
>>> wrote:
>>>
>>>> The problem is in the packages loaded (or not) by Sage's GAP vs by
>>>> GAP's GAP.
>>>> Namely, if I start Sage's gap by ./sage --gap
>>>> and do
>>>> gap> LoadAllPackages();
>>>> gap> NormalSubgroups(SmallGroup(1458,1180));
>>>> I'd get your error
>>>>
>>>> Now, if I take GAP's source for 4.8.6:
>>>>
>>>> https://www.gap-system.org/pub/gap/gap48/tar.bz2/gap4r8p6_2016_11_12-14_25.tar.bz2
>>>> untar it, and copy its pkg/ directory to $SAGEROOT/local/gap/latest/
>>>> start Sage's gap by ./sage --gap
>>>> and do
>>>> gap> LoadAllPackages();
>>>> gap> NormalSubgroups(SmallGroup(1458,1180));
>>>> [ <pc group of size 1458 with 7 generators>, Group([ f1, f2, f3^2*f4,
>>>> f5, f6, f7 ]), Group([ f2, f3^2*f4, f4, f5, f6, f7 ]),
>>>> Group([ f1, f3^2*f4, f4, f5, f6, f7 ]), Group([ f1, f2*f4*f6,
>>>> f3^2*f4, f5, f6, f7 ]), Group([ f1, f2^2*f4*f6^2, f3^2*f4, f5, f6, f7 ]),
>>>> Group([ f1, f3^2*f4, f5, f6, f7 ]), Group([ f3^2*f4, f4, f5, f6, f7
>>>> ]), Group([ f2*f4*f6, f3^2*f4, f5, f6, f7 ]),
>>>> Group([ f2^2*f4*f6^2, f3^2*f4, f5, f6, f7 ]), Group([ f2, f3^2*f4,
>>>> f5, f6, f7 ]), Group([ f1, f3^2*f4, f5*f6^2, f7 ]), Group([ f3^2*f4, f5,
>>>> f6, f7 ]),
>>>> Group([ f2*f3^2*f4, f3^2*f4^2, f5*f6^2, f6, f7 ]), Group([ f3^2*f4^2,
>>>> f5*f6^2, f6, f7 ]), Group([ f2*f3*f6*f7, f5*f6^2, f6, f7 ]),
>>>> Group([ f2^2*f4*f6^2, f5*f6^2, f6, f7 ]), Group([ f2*f3^2*f4,
>>>> f5*f6^2, f6, f7 ]), Group([ f5*f6^2, f6, f7 ]), Group([ f3^2*f4, f5*f6^2,
>>>> f7 ]),
>>>> Group([ f3^2*f4^2, f5^2*f6^2, f7 ]), Group([ f5^2*f6^2, f7 ]),
>>>> Group([ f5*f6^2, f7 ]), Group([ f7 ]), Group([ ]) ]
>>>>
>>>> --- it works.
>>>> Note that LoadAllPackages() is necessary, as apparently the GAP code
>>>> uses one of these packages for these
>>>> computation without telling us.
>>>>
>>>> So this is a sort of bug, yes, in our way of packaging GAP packages
>>>> (and it's a sort of GAP's bug to call a code from a package
>>>> without loading it explicitly...)
>>>> I don't know what package it is yet.
>>>> It's very early Sat. morning here, I'll have a look at it over the
>>>> weekend.
>>>>
>>>>
>>>> On Saturday, February 17, 2018 at 3:12:25 AM UTC, David Roe wrote:
>>>>>
>>>>> I'm trying to use GAP's small groups library (after installing
>>>>> gap_packages) and getting strange errors that don't occur in GAP when
>>>>> built
>>>>> from source. In particular (using gap_console to get more traceback)
>>>>>
>>>>> sage: gap_console()
>>>>> ...
>>>>> gap> NormalSubgroups(SmallGroup(1458,1180));
>>>>> Error, List Element: <list>[4] must have an assigned value in
>>>>> if map[d] > max then
>>>>> elm := id;
>>>>> else
>>>>> if not IsBound( map[d] ) then
>>>>> Error( "<elm> lies not in group defined by <pcgs>" );
>>>>> fi;
>>>>> ll := LeadingExponentOfPcElement( pa, elm );
>>>>> lr := lc[d];
>>>>> exp[map[d]] := ll / lr mod ros[d];
>>>>> elm := LeftQuotientPowerPcgsElement( pcgs, map[d], exp[map[d]],
>>>>> elm );
>>>>> fi; at /home/roed/sage/local/gap/latest/lib/pcgsind.gi:1411 called
>>>>> from
>>>>> ExponentsOfPcElement( NumeratorOfModuloPcgs( pcgs ), elm,
>>>>> pcgs!.depthMap ) at /home/roed/sage/local/gap/latest/lib/
>>>>> pcgsmodu.gi:931 called from
>>>>> ExponentsOfPcElement( ocr.modulePcgs, x ) at
>>>>> /home/roed/sage/local/gap/latest/lib/onecohom.gi:264 called from
>>>>> ocr.moduleMap( n ) at /home/roed/sage/local/gap/latest/lib/
>>>>> onecohom.gi:339 called from
>>>>> ocr.listToCocycle( lc ) at /home/roed/sage/local/gap/latest/lib/
>>>>> grplatt.gi:1926 called from
>>>>> opr( Zero( vs ), k ) at /home/roed/sage/local/gap/latest/lib/
>>>>> grplatt.gi:1932 called from
>>>>> ... at line 2 of *stdin*
>>>>>
>>>>> Similar problems are occurring with other groups and some other
>>>>> functions.
>>>>>
>>>>> In the interest of finishing my calculation, I tried the following in
>>>>> an attempt to have Sage use a functional GAP:
>>>>>
>>>>> sage: import sage.interfaces.gap
>>>>> sage: sage.interfaces.gap.gap_cmd = "/home/roed/gap4r8/bin/gap.sh"
>>>>> sage: gap.eval('1+1')
>>>>>
>>>>> which hangs. Similarly, I tried copying my functional gap
>>>>> installation to $SAGE_LOCAL/gap/ and updating the "latest" symlink; this
>>>>> allows gap_console() to function correctly but gap.eval() still hangs.
>>>>>
>>>>> I don't tend to work with spkgs very often, so I feel like there's
>>>>> probably something simple I'm overlooking. Any help in installing a
>>>>> functioning gap would be appreciated. I'm also happy to help investigate
>>>>> what's wrong with our process for installing the small groups library.
>>>>> David
>>>>>
>>>> --
>>>> You received this message because you are subscribed to the Google
>>>> Groups "sage-devel" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send
>>>> an email to sage-devel+...@googlegroups.com.
>>>> To post to this group, send email to sage-...@googlegroups.com.
>>>> Visit this group at https://groups.google.com/group/sage-devel.
>>>> For more options, visit https://groups.google.com/d/optout.
>>>>
>>>
>>>
>>
--
You received this message because you are subscribed to the Google Groups
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to sage-devel+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.
