py3: hash for projective space

[chapoton]

part of #24551

[chapoton]

New commits:

​9f685f3

py3: adding hash for projective spaces

[git]

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

​2caa7ae

fixing one 64bits hash

[chapoton]

@tmonteil, could you please help me fix the 32 bits hash, by running the following doctest with this branch on your 32 bit machine and reporting the result ?

            sage: P.<x,y> = ProjectiveSpace(Zmod(10), 1)
            sage: hash(P([2, 5]))
            -479010389                     # 32-bit

[tmonteil]

I stopped my patchbot and restarted it on this ticket, see ​https://patchbot.sagemath.org/log/Pending/26091/debian/9.4/i686/4.9.0-6-686/tmonteil-debian-stretch-32/2018-08-19%2011:15:44

[tmonteil]

And the result is:

sage -t --long src/sage/schemes/projective/projective_point.py
**********************************************************************
File "src/sage/schemes/projective/projective_point.py", line 402, in sage.schemes.projective.projective_point.SchemeMorphism_point_projective_ring.__hash__
Failed example:
    hash(P([2, 5]))
Expected:
    -479010389                     
Got:
    -2047536788
**********************************************************************

[git]

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

​eafa8b6

fix hash for 32bit

[chapoton]

Merci beaucoup !

[tscrim]

Note that #23807 plans to solve this in a different way by making projective space use UniqueRepresentation.

[saraedum]

Would you mind replacing the sage: hash(P([2, 5])) with sage: hash(P([2, 5])) == hash(P([2, 5]))? I find these 32/64 bit tests really annoying because you cannot update these easily.

[saraedum]

If you changed that, it would be a positive review from me. But I haven't looked at #23807, so I leave it to you to figure out whether that one should go in instead.

[tscrim]

I expect this to be done and merged before #23807.

[git]

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

​a3bddad

detail in #26091 (hash doctest)

[chapoton]

done (note that this is a case where the hash does not depend on the point..)

[chapoton]

Everything seems to be ok. Positive review, somebody ?

[chapoton]

review, please

[chapoton]

should we wait for #23807 ?

[tscrim]

No, let's just get this in and rebase #23807 on this.

[embray]

Ah, I just noticed this one. Thanks. I was just starting to fix this myself but then I remembered to check first (actually I had already added a hash for this class a while ago, but only as a quick hack--never got around to documenting it or anything, so this is better).