OEIS update (database format change, new entries, incorrect warning handling)

[Sébastien Labbé]

This ticket fixes the 34 doctest failures that happen when internet is enabled, see ​this sage-release thread.

There are 3 kinds or errors:

• the oeis database gets updated with new data (this is a normal update),
• the oeis database changed its internal format, and does not make difference between signed and unsigned sequences anymore (in particular, for signed sequences, the V,W,X fields disapear, and are replaced with the S,T,U fields),
• the warnings triggered when a dead sequence is met were not correctly doctested.

The 3 commits corresponds to these 3 kind of errors

To test the branch:

sage -t --optional=sage,internet src/sage/databases/oeis.py 

[Sébastien Labbé]

shorter title

[Thierry Monteil]

Hi,

i will be on this (some might be tricky, e.g. related to dead sequences).

You write:

Note that, as Sloane writes this week on the ​arxiv, the oeis is a now a Wiki which means that Sage doctests should depend as few as possible on the many modifications done on the OEIS database...

Are you suggesting to parse the wiki ???

Note that the issue of the database changing was discussed in previous tickets. Still, we should show part of the answers for having meaningful examples (not only for testing purpose, but also for the user).

[John Palmieri]

By the way, several of us predicted this sort of situation (lots of broken doctests related to OEIS) in #16252.

[git]

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

​cf7a0aa

merge

[Thierry Monteil]

New commits:

​e081bc5	Trac 25471 : fix the 20 trivial upstream changes
​6f15c06	Trac 25471 : the OEIS database format changed and does not make difference between signed and unsigned sequences anymore
​9889222	Trac 25471 : fix warnings for dead sequences

[Frédéric Chapoton]

ok (but this will need to be done again and again...)