Re: abysmal check-portability speed

[Jonathan Perkin <jperkin%mnx.io@localhost>]

* On 2024-09-23 at 16:44 BST, Greg Troxel wrote:

> Jonathan Perkin <jperkin%mnx.io@localhost> writes:
>
> > You might find this patch helps:
> >
> >   https://github.com/TritonDataCenter/pkgsrc/commit/21aa3cb38903181f291ceb5ccb0245af12fe7a00
>
> Thanks - that is a huge clue to what's going on.
>
> It seems that NetBSD's /bin/sh does not support read -n.

For the interest of others, I've pushed a commit to mktool to add a 
check-portability replacement.  As expected it's significantly faster 
(~16x, 3.3s vs 52.1s), testing in x11/qt5-qtwebengine:

  https://github.com/jperkin/mktool/commit/fdcebff821943bc8500eaccb635da62dc9d85b0e

While developing this I realised that the "read -n" patch is harmful, at 
least with certain shells:

  $ printf "foo\nbar\n" | bash -c 'read -n 128 foo; echo $foo'
  foo

  $ printf "foo\nbar\n" | mksh -c 'read -n 128 foo; echo $foo'
  foo bar

This actually breaks the tests completely when using mksh, as it no 
longer matches "#!*/bin/sh" as the first line.  So, something to watch 
out for.

mktool users: if you're able to give that patch a spin and let me know 
how it goes I'd appreciate it.  I will get a new version in after the 
freeze.

--
Jonathan Perkin   -   mnx.io   -   pkgsrc.smartos.org
Open Source Complete Cloud   www.tritondatacenter.com