FIX: Maximal subgroups will work also if no tomlib available

[fingolfin]

This extracts a commit from PR #3494 by @hulpke

Fixes #3498

[fingolfin]

Perhaps this also fixes #3496?

[olexandr-konovalov]

One manual example is affected: https://travis-ci.org/gap-system/gap/jobs/545707527. A fix is part of commit c4116b0. The fix seems to work - tested it on manual examples and teststandard, those do not timeout any more.

A problem from #3496

########> Diff in /Users/alexk/GITREPS/gap/tst/testbugfix/2019-04-09-Lattice.t\
st:3
# Input is:
StructureDescription(g);;t:=TableOfMarks(g);;
# Expected output:
# But found:
Error, List Element: <list>[11] must have an assigned value
########

when all packages are loaded still stays - that one should be fixed elsewhere in #3494.

[olexandr-konovalov]

This fix will still break testextra, but at least will not cause timeout and may be fixed later:

########> Diff in /Users/alexk/GITREPS/gap/tst/testextra/grpperm.tst:266
# Input is:
p:=List(p,x->Image(IsomorphismPermGroup(x)));
# Expected output:
[ <permutation group with 13 generators>,
  <permutation group with 13 generators>,
  <permutation group with 13 generators>,
  <permutation group with 13 generators> ]
# But found:
[ <permutation group with 13 generators>, 
  <permutation group with 13 generators>, 
  <permutation group with 13 generators>, 
  <permutation group with 13 generators> ]
########

Look identical, eh? The diff is in a space after the comma. Cf. the idea of comparing output up to trailing whitespaces from #697.

[codecov]

Codecov Report

Merging #3501 into master will decrease coverage by 0.06%.
The diff coverage is 16.66%.

@@            Coverage Diff             @@
##           master    #3501      +/-   ##
==========================================
- Coverage   85.43%   85.36%   -0.07%     
==========================================
  Files         699      699              
  Lines      346760   346766       +6     
==========================================
- Hits       296253   296025     -228     
- Misses      50507    50741     +234

Impacted Files	Coverage Δ
lib/twocohom.gd	100% <ø> (ø)	⬆️
lib/maxsub.gi	66.57% <0%> (-0.46%)	⬇️
lib/grplatt.gi	78.31% <100%> (ø)	⬆️
lib/process.gi	36% <0%> (-8%)	⬇️
lib/float.gd	96.77% <0%> (-3.23%)	⬇️
lib/info.gi	82.53% <0%> (-2.63%)	⬇️
lib/grpfp.gi	79.39% <0%> (-2.07%)	⬇️
lib/relation.gi	71.7% <0%> (-1.91%)	⬇️
lib/dict.gi	77.47% <0%> (-1.27%)	⬇️
lib/magma.gi	89.11% <0%> (-0.96%)	⬇️
... and 20 more

[coveralls]

Coverage decreased (-0.002%) to 85.256% when pulling f9eeeb9 on fingolfin:mh/fix-fr into ff93de2 on gap-system:master.

[hulpke]

Note that -- as requested -- I now review what is de facto my own code.
It would be nice if #3494 also could be approved, as waiting for the formal approval takes resources.

[fingolfin]

Are you sure that breakage in tst/testextra/grpperm.tst:266 is caused by this PR, and not rather by some other change, and only uncovered by this PR fixing the timeout in tst/testextra/grpperm.tst? Because I don't see how this PR could possible cause that diff.

Those spaces at the end of the lines really should be there, so it seems more likely to me that this was already broken when the test was added in commit 97711e4 / PR #3451. I've now fixed this via another commit in this PR.

[olexandr-konovalov]

Thanks @fingolfin !

[olexandr-konovalov]

And yes, I agree with your analysis @fingolfin (hence rebase and merge is OK for this PR). When the test run out of time, tst/testextra/grpperm.tst was never reached at all.