Adding a new collection to FindStat breaks the interface

[stumpc5]

The interface is now broken due to the newly added collection. There should be a way to avoid this break.

CC: @mantepse

Component: interfaces

Keywords: FindStat

Author: Martin Rubey

Branch/Commit: u/mantepse/develop @ 151934e

Reviewer: Christian Stump

Issue created by migration from https://trac.sagemath.org/ticket/19543

[mantepse]

Branch: u/mantepse/develop

[mantepse]

Commit: 3c17593

[stumpc5]

comment:3

Looks good to me -- do you also want to use this ticket to add the support of binary words, or do you want to create another?

[mantepse]

comment:4

I thought of putting the binary words in another ticket, because otherwise this one cannot be tested...

[stumpc5]

comment:5

I am not sure whether that's the right way to go -- e.g., currently all doctests of findstat.py fail due to this print.

[stumpc5]

Reviewer: Christian Stump

[stumpc5]

Author: Martin Rubey

[stumpc5]

comment:7

Here is another problem to fix:

sage: findstat({l:l.length() for l in Partitions(4)})

---------------------------------------------------------------------------
IOError                                   Traceback (most recent call last)
...
IOError: FindStat did not answer with a json response: Could not find FindStat collection for Binary words.

Before, we had the problem that FindStat was broken, so the thrown error was

IOError: FindStat did not answer with a json response: St000288 is not a FindStat statistic identifier.

I think that the finder functionality should never throw an error if one of the results is corrupted in any way (this way, the interface is robust even if FindStat is broken).

One quick idea: what about having a method findstat.collections such that

sage: findstat.collections()
{'Binary words': 'not supported',
 'Dyck paths': 'supported',
 'Permutations': 'supported'}

and if an unsupported collection is used, we throw an error like the one you have.
Good:

sage: findstat(289)
ValueError: Could not find FindStat collection for Binary words.

but unclear what to do in the above situation where a binary word statistic is found and we cannot throw an error...

[mantepse]

comment:8

Oh no! Thanks for pointing this out! Almost a fix is committed.

[sagetrac-git]

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

151934e

almost a fix

[sagetrac-git]

Changed commit from 3c17593 to 151934e

[mantepse]

comment:10

This is subsumed by #19296.

[fchapoton]

comment:12

can we close that now ?

[stumpc5]

comment:14

I believe so. @mantepse ?

[mantepse]

comment:15

Yes, but I don't know how to do this. I set it to wontfix, is this not enough?

[fchapoton]

comment:16

you need to set to positive_review, then the release manager will close