Deprecation warnings warning

[Rob Beezer]

Mostly for Andrew.

I keep seeing a warning that we are only checking two years worth of
deprecations, even with very minor or trivial stylesheets. So I thought I would
stop doing any checking at all for these minor stylesheets. Turns out we are
not doing this checking - just warning.

There is a global variable being set (two years worth, or all of them) and a
warning is being issued "globally" as soon as the variable is set, not when
employed/consulted. For truth-in-advertising, I can move the warning to where
the checks are *really* happening. But before I do that, I want to see how we
feel about this option in the first place (two years v. all).

There is a lot of nice code to make the warnings only "work" for those that are
currently less than two years old. But I think we decided in the end that we
had been sort of fooled by how much of a performance hit these were. And it was
really only a small fraction of total run-time, even with all the speed-ups
we've made in the past year or so.

In the name of simplicity, do we want to back out the option to only look back
two years? Andrew - might you be able to do some updated profiling for LaTeX
and HTML? The relevant templates are some subset of those with modes:

"generic-warnings"
"element-deprecation-warnings"
"parameter-deprecations-warnings"

Another option might be expanding the look-back time (four years?) as the pace
of deprecations has slowed?

Thanks,
Rob

[Andrew Scholer]

I just reran timings on a couple of books. 

For a full build of decent size book, those account for ~0.5% of the build when set to full checks. About 0.07% for 2 years.

For a subtree build, that climbs to about 3% for full checks or 0.4% for 2 year.

So yes, currently not a huge deal. But the 3% for subtree build is not nothing. One of the takeaways from the efficiency work should be that a bunch of small speedups (or slowdowns) add up. There were a few big ticket items, but many of the individual changes were pretty small on their own.

Maybe we move the message and change the default to all (the 100 years)?

Andrew Scholer (he/him/his)

Computer Science Instructor

Chemeketa Community College

503.589.7649 

computerscience.chemeketa.edu/people/andrew-scholer/

--
You received this message because you are subscribed to the Google Groups "PreTeXt development" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pretext-dev...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/pretext-dev/MTAwMDAxOS5iZWV6ZXI.1756169877%40pnsh.