#27830 closed enhancement (fixed)
py3: fix remaining doctests for groups/perm_gps
Reported by:  jhpalmieri  Owned by:  

Priority:  major  Milestone:  sage8.8 
Component:  python3  Keywords:  
Cc:  Merged in:  
Authors:  John Palmieri  Reviewers:  Travis Scrimshaw 
Report Upstream:  N/A  Work issues:  
Branch:  c5b7062 (Commits)  Commit:  
Dependencies:  Stopgaps: 
Description (last modified by )
Fix all Python 3 doctests in groups/perm_gps
.
Change History (16)
comment:1 Changed 5 months ago by
 Branch set to u/jhpalmieri/py3groups
comment:2 Changed 5 months ago by
 Commit set to 402d35631cdcc8a685cf4cd492a6bc3d629b4219
 Status changed from new to needs_review
comment:3 Changed 5 months ago by
 Commit changed from 402d35631cdcc8a685cf4cd492a6bc3d629b4219 to c155241de8c2ab86769fe0ec4739433e8b427e69
Branch pushed to git repo; I updated commit sha1. New commits:
c155241  trac 27830: last fix in groups/perm_gps

comment:4 Changed 5 months ago by
 Description modified (diff)
Wait, I do know how to fix that last failure.
comment:5 Changed 5 months ago by
 Summary changed from py3: some fixes for groups/perm_gps to py3: fix remaining doctests for groups/perm_gps
comment:6 Changed 5 months ago by
How about if the sorting fails, you run a sorted(foo, key=str)
instead so it becomes deterministic?
comment:7 Changed 5 months ago by
As Jeroen pointed out at #26966, str
is not an injective function:
sage: L1 = [1.0 + 2^52, 1.0]; L2 = reversed(L1) sage: sorted(L1, key=str) == sorted(L2, key=str) False
So it won't be deterministic even with key=str
.
comment:8 Changed 5 months ago by
Ah, right. Although I guess the weaker form of the general question would be should we try to make it as deterministic as possible?
comment:9 Changed 5 months ago by
I don't have strong feelings about it. Does it matter if it produces different results on Python 2 vs. Python 3?
comment:10 Changed 5 months ago by
I don't think so (beyond the doctests). I tend to enjoy more deterministic results as it makes it easier to debug and write doctests.
comment:11 Changed 5 months ago by
 Commit changed from c155241de8c2ab86769fe0ec4739433e8b427e69 to c5b7062cacbab409a0502509e9917fe5d9946a4a
Branch pushed to git repo; I updated commit sha1. New commits:
c5b7062  trac 27830: make things a bit more deterministic by sorting with key=str

comment:12 Changed 5 months ago by
Here's a more deterministic version.
comment:13 Changed 5 months ago by
 Reviewers set to Travis Scrimshaw
 Status changed from needs_review to positive_review
Great, thank you.
comment:14 Changed 5 months ago by
 Branch changed from u/jhpalmieri/py3groups to c5b7062cacbab409a0502509e9917fe5d9946a4a
 Resolution set to fixed
 Status changed from positive_review to closed
comment:15 Changed 5 months ago by
 Commit c5b7062cacbab409a0502509e9917fe5d9946a4a deleted
I think this caused some breakage at #27863
comment:16 Changed 5 months ago by
I don't see anything in the changes here that would make any difference with Python 2 behavior.
There is one remaining failure:
I don't know how to fix this.
New commits:
trac 27830: fix most py3 doctests in groups/perm_gps