Re: Parallelization of syscmd

On Thu, Sep 18, 2025 at 06:37:38PM -0300, Bento Borges Schirmer wrote:
> Hey,
> 
> I don't really know how m4 works. That's just an idea that occurred to
> me. I think m4 could keep expanding macros and processing text while a
> syscmd() is in flight, potentially having several syscmd() executing
> at once. The user would have to guarantee that the syscmd() execution
> is "pure", that is, that one command does not depend on side effects
> of previous executions. All that could be enabled by some flag, like
> -j which reminisces Makefile.

You are welcome to try and write such a patch.  But currently m4 is
single-threaded; what you are proposing would be a massive redesign,
in order to handle parallel threads and possible speculative execution
followed by rollback if the syscmd was not pure after all.

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.
Virtualization:  qemu.org | libguestfs.org