Doctest failure in LocalMcLaughlinGraph

[jdemeyer]

Obvious failure introduced in #18782:

**********************************************************************
File "src/sage/graphs/generators/smallgraphs.py", line 3453, in sage.graphs.generators.smallgraphs.LocalMcLaughlinGraph
Failed example:
    g = graphs.LocalMcLaughlinGraph(); g   # long time # optional - gap_packages
Expected nothing
Got:
    Local McLaughlin Graph: Graph on 162 vertices
**********************************************************************

[ncohen]

Sigh... sage -btp insead of sage -btp -l.. Sorry :-/

Nathann

P.S.: Now a broken optional doctest is a blocker? Cool :-P

[git]

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

​05a2e41

trac #18911: Broken doctest

[jdemeyer]

Replying to ncohen:

Now a broken optional doctest is a blocker? Cool :-P

Every doctest failure should be a blocker. That's why I care so much to not introduce artificial doctest failures in #18904.

[ncohen]

Every doctest failure should be a blocker.

It's a bit weird to say that something is a blocker even though our automated testing (patchbots, Volker's tests before each release) do not detect it.

That's why I care so much to not introduce artificial doctest failures in #18904.

If only you did not try to hid bugs under the carpet in order to preserve the appearance of reliability ...

I also want all doctests to pass. My method to do so is to let them scream when they want to, and to fix the problems.

Nathann

[ncohen]

By the way, given that 'every doctest failure should be a blocker', why don't you apply that and create blocker tickets until all broken doctests involving maple/matlab/octave/mathematica are fixed? Or is it actually a lie? :-P

[jdemeyer]

Replying to ncohen:

It's a bit weird to say that something is a blocker even though our automated testing (patchbots, Volker's tests before each release) does not even detect it.

That is certainly true. Better testing infrastructure is hopefully something that will happen with support from OpenDreamKit.

[ncohen]

That is certainly true. Better testing infrastructure is hopefully something that will happen with support from OpenDreamKit.

Indeed. Well if you ever meet those guys please tell them about it. That would definitely help.

[jdemeyer]

Replying to ncohen:

why don't you apply that and create blocker tickets until all broken doctests involving maple/matlab/octave/mathematica are fixed?

I don't see any such failures in ptestlong.log, they are all under my big carpet.

[ncohen]

I don't see any such failures in ptestlong.log, they are all under my big carpet.

Sigh... Well, the good thing about this bug is that it says that there are reasons for me to re-run my patchbot again. With many packages.

Nathann

[ncohen]

... seems that there was a new power cut in my office in Paris. Okayyyyy, another time :-P

[ncohen]

Thaaaaaaaaaaaaaaaaaaaaaanks,

Nathann