On Fri, 20 May 2022 at 10:03, Jonathan Wakely <jwak...@redhat.com> wrote:
>
> On Fri, 20 May 2022 at 03:20, Hans-Peter Nilsson via Libstdc++
> <libstd...@gcc.gnu.org> wrote:
> >
> > Also, how about shortening those test-suite file-paths?
> > They're path sort-of overlong for any git (and ChangeLog)
> > commit-line limit. 1/2 :-)
>
> Yes, they're silly. I like what libc++ does, i.e. name the testsuite
> directories after the [stable.name] tags in the standard. They have
> std/rand/rand.dis/rand.dist.bern/rand.dist.bern.negbin (for some
> reason they use "rand.dis" instead of "rand.dist" for the third
> component there, which is not what the standard has,
> https://eel.is/c++draft/rand.dist.bern.negbin shows the subclause
> hierarchy).
>
> That's still quite unwieldy, so we could remove the repetition and use
> 26_numerics/rand/dist/bernoulli/negbin/. The problem now is that it
> gets quite cryptic, so finding the tests for the
> negative_binomial_distribution is a bit harder if you don't know where
> to look (but git grep solves that in less than a second).

Or we could forget about mirroring the nesting in the standard and
just use 26_numerics/random/dist/negative_binomial

That's still too long for ChangeLogs though.

Reply via email to