Don't require matplotlib or sympy to run doctests

[jdemeyer]

The doctest framework itself should not depend on matplotlib or sympy. This is because some other packages (p_group_cohomology for example) use the Sage doctest framework but do not otherwise require those packages.

[git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

​058ae5f

Don't require matplotlib or sympy to run doctests

[jhpalmieri]

This looks okay to me. Simon, any opinions?

[SimonKing]

Replying to jhpalmieri:

This looks okay to me. Simon, any opinions?

I'd find it very good if one could use the SageMath doctest framework with as few prerequisites as possible. So, if this branch makes it so that sympy and matplotlib are not required to run doctests, I am +1!

[SimonKing]

On the other hand: Both sympy and matplotlib are standard packages, and I don't see what harm would be done if Sage requires two of its standard packages to run tests.

[jdemeyer]

Replying to SimonKing:

I don't see what harm would be done if Sage requires two of its standard packages to run tests.

There is harm if the Sage doctesting framework requires two of Sage's standard packages to run tests because the Sage doctesting framework is used for non-Sage tests too in #18514.

[SimonKing]

Replying to jdemeyer:

Replying to SimonKing:

I don't see what harm would be done if Sage requires two of its standard packages to run tests.

There is harm if the Sage doctesting framework requires two of Sage's standard packages to run tests because the Sage doctesting framework is used for non-Sage tests too in #18514.

No it isn't. The p_group_cohomology spkg has always and foremost been a Sage package (that's what spkg stands for). So, the p_group_cohomology's test suite requires Sage being present. Otherwise it would simply not run.

Best regards, Simon

[jdemeyer]

This ticket is not about p_group_cohomology. I don't think that the Sage doctesting framework should require all standard Sage packages.

[jhpalmieri]

I'm willing to give this a positive review. I don't see a downside to it, and there are potential advantages, even if there are no immediate uses for it yet (#18514 doesn't require it, I think).

[saraedum]

Looks good to me. jhpalmieri, feel free to add yourself to the reviewer list.

jdemeyer: So, our doctesting framework could/should be an independent standard SPKG then?

[jdemeyer]

Replying to saraedum:

jdemeyer: So, our doctesting framework could/should be an independent standard SPKG then?

Depends what you mean with "independent". It should certainly be usable by other packages besides Sage (for example p_group_cohomology). But it still depends on pieces of Sage, so it's not that simple to make it truly independent.