graph_plot: sort by id

[jdemeyer]

Avoid using comparison (which may not always work, may not provide a total order, etc.) when all we want as far (as I understand) is an arbitrary total order.

CC: @mezzarobba

Component: graph theory

Author: Marc Mezzarobba

Branch/Commit: u/jdemeyer/graph_plot__sort_by_id @ 983e660

Issue created by migration from https://trac.sagemath.org/ticket/22375

[jdemeyer]

Branch: u/jdemeyer/graph_plot__sort_by_id

[mezzarobba]

Commit: 983e660

[mezzarobba]

New commits:

983e660

graph_plot: sort by id

[jdemeyer]

comment:3

I don't get the point of this ticket. There are so many places where G.vertices() is called for a graph G. Why are all those other places ok but not this specific one that you are changing in this ticket?

[mezzarobba]

comment:4

Replying to @jdemeyer:

I don't get the point of this ticket. There are so many places where G.vertices() is called for a graph G. Why are all those other places ok but not this specific one that you are changing in this ticket?

I don't really remember. But looking at the commit, it looks like this code wants to consider edges (u,v) with u ≤ v in the order in which the vertices are looped over. Sorting them by id is an easy way to avoid doing true comparisons without making deeper changes...

[jdemeyer]

comment:6

Replying to @mezzarobba:

I don't really remember.

So you're asking to review a ticket when even the ticket author cannot remember the purpose of that ticket?

[mezzarobba]

comment:7

Replying to @jdemeyer:

Replying to @mezzarobba:

I don't really remember.

So you're asking to review a ticket when even the ticket author cannot remember the purpose of that ticket?

I do remember that the goal was to fix problems that arose when trying to disable comparisons between objects with incompatible parents (#22029)—as you know, since it was you who wanted to make this a separate ticket. I don't remember exactly why I thought this would be a correct fix here but not elsewhere. And I'm hoping for comments from people who know this code better than I do, hence the needs_review.

[jdemeyer]

comment:8

As long as you don't expect me to review it :-)

[fchapoton]

comment:9

does not apply

[jdemeyer]

comment:10

Any objections to closing this as "wontfix"?

[fchapoton]

comment:11

no objection from me