On Sep 11, 2025, at 9:36 PM, Michael Paquier <michael@paquier.xyz> wrote:
On Thu, Sep 11, 2025 at 06:56:07AM -0400, Greg Burd wrote:
Just for reference I started this not to increase coverage, which is a good
goal just not the one I had. I was reviewing the API and considering some
changes based on other work I've done. Now that I see how deeply baked in
this code is I think that's unlikely. Maybe something else distinct for
bitmaps over 64-bit space at some point will be useful. I wrote this code
just to capture the API in test form.
How much does this measure in terms of numbers produced by
coverage-html (see [1])? The paths taken don't always matter as it
can also be important to check combinations of code paths that are
taken by other tests when checking after edge cases, but that would
give an idea of gain vs extra runtime. Not objecting to your patch,
just being curious as I am not seeing any numbers posted on this
thread.
[1]: https://www.postgresql.org/docs/devel/regress-coverage.html
--
Michael
Sawada-san, Michael,
Thank you both for the push to measure. Before the patch as it stands now the
coverage for src/backend/nodes/bitmapset.c is 63.5% and after it is 66.5%. Not
an amazing difference, but something. I guess I expected this to be higher given
the degree to which this datatype is used.
I'll review the gaps in coverage and update the tests. I'll look for a way to add
meaningful randomization.
-greg
