Re: tools/isl won't build (GCC 12??)

[<tlaronde%kergis.com@localhost>]

On Sun, Jun 16, 2024 at 12:50:52PM -0400, Mouse wrote:
> >> I'm trying to build -current and tools/isl is failing [...]
> > thanks, these problems are usually found randomly like this which is
> > unfortunate.
> 
> I have patches to (5.2) touch which are designed to facilitate such
> things.  To randomly reorder the modtimes under /usr/src:
> 
> find /usr/src -type f | shuffle -f - | touch -d -1 -i - -t `date +%Y%m%d%H%M.%S`
> 
> -d and -i are the additions (5.2 predates -current's -d):
> 
>      -d inc  Specify inc, a time increment (`d'elta) in seconds, which is
>              added to the time used for each file to give the time used for
>              the next file; other options which control the timestamp to be
>              used specify the time for the first-affected file.  This has no
>              effect unless multiple files are affected.  inc must be an inte-
>              ger, and may be negative (producing progressively earlier times).
> 
> and
> 
>      -i      Indirect: any file arguments are not files to be manipulated;
>              rather, they name files containing filenames, one per line, which
>              are to be manipulated.  As a special case, an argument of `-'
>              causes names to be read from the standard input.  (This special
>              case does not apply unless -i is given.)  This is somewhat like
>              using xargs(1) with touch; this option exists because (a) it is
>              necessary to allow -d to operate properly even on very long lists
>              of files and (b) xargs(1) considers space and tab, as well as
>              newline, to delimit arguments.
> 
> I can supply diffs relative to 5.2, which may be a useful starting
> point if anyone wants to adapt these to -current's touch(1).

FWIW, it seems that SOURCE_DATE_EPOCH is used for example when
generating python byte-compiled version of scripts, in order for
independent compilations (typically C.I. and pipelines) to yiel the
same results.

Maybe this could be used if set in the environment as date?

: ${SOURCE_DATE_EPOCH:=$(date "+%FT%T")}

and using '-d' with touch(1) instead of '-t'.

FWIW, I stumbled about all this when working with Xorg modules, Xorg
going from Autotools to Meson as compilation/installation framework .

Doc is here:

http://notes.kergis.com/x11_building.html

(The aim being to have userland X11 build without ado under NetBSD, in
order to put the whole stuff hopefully out of src and out of problem
source for GPU DRM woes. Note: I use neither Autotools nor Meson, and
even not Python; but the meson.build is more readable and in my
work---see above---I wrote POSIX.2 script to handle Python so that not
Meson users can translate building easily.)
-- 
        Thierry Laronde <tlaronde +AT+ kergis +dot+ com>
                     http://www.kergis.com/
                    http://kertex.kergis.com/
                     http://nunc-et-hic.fr/
Key fingerprint = 0FF7 E906 FBAF FE95 FD89  250D 52B1 AE95 6006 F40C