[OpenIndiana-discuss] on shell/fish patches

[Paolo Vincenzo Olivo]

On 25/11/08 08:31PM, Geoff Weiss wrote:
> > I do recall now that when I attempted to run gmake
> > test, there was an error at some point that said something like unable
> > to determine file type.  The 02 patch had file type related stuff in it
> > so I ported it over to rust.  It didn't fix the problem.  I handed the
> > recipe over to Till since he know rust.  It languished for months.
> > After several pokes, he came back and said that patches were good.  The
> > file type detection code didn't get fixed. By that time, I no longer
> > knew the history of what I had did.  So it got distributed as-is.
> > 
> > I eventually got gmake test to start with fish 4.1.2.  I am not
> > bothering trying to run tests in the 4.0.9 PR that I just sent to
> > OpenIndiana today. gmake test in 4.1.2 stops in the middle due to an
> > error.
> >

> 
> One clarification.  The issue that I thought was a file type detection error
> when running tests with fish 4.0.x was not due a fish compilation error.
> There was a problem where files were not being written to the correct
> directory.  The problem does not happen with fish 4.1.x.
>

Thanks for the thorough explanation! 
Hope I can get fixes upstreamed in the next fish release.
Rust lang libc/nix patches will require a different approach though.

The pkgsrc package for shells/fish4 also adds a fix for SunOS
msgfmt(1), [0] which does not support a `--check-format' flag.

[0] https://github.com/NetBSD/pkgsrc/blob/trunk/shells/fish4/patches/patch-crates_gettext-maps_build.rs

>  Please let me know if you can find somebody who is willing to fix the
>  code in src/wutil/dir_iter.rs so the filetype detection can work on
>  Solaris/Illumos.

Sure, this is something I'm also interested in working out. I have some
patches on pkgsrc which provide a workaround for d_type in struct
dirent. See for example net/telescope.[1] However, lacking any kind of
experience with Rust, I have to investigate this further before coming
up with something safe committing (and a review from someone more
knowledgeable would be required either way).

In case, I will post a follow-up on this thread.

[1] https://github.com/NetBSD/pkgsrc/blob/trunk/net/telescope/patches/patch-fs.c

--
PVO