On Sun, 2026-04-19 at 21:58 +0530, Shubham Chakraborty wrote:
> Thanks for your feedback
> 
> I'll drop the incremental patch approach and instead look into moving
> this toward the gnulib-port integration.
> I will make sure that any work I submit is based on the existing GNU
> Make and Gnulib sources and written manually, following project
> guidelines.

I think this will open up a large can of worms.

I am not willing to give up the bootstrapping script.  Just last week I
got a report from someone who wanted to use it, so I don't think that
it's unreasonable to assume it's still important in some contexts.

Also it's not only the bootstrapping but also the portability to
Windows and VMS that's an issue.

The problem with using more gnulib is that gnulib modules are more and
more interdependent: you can't pull in any one module without also
requiring many other modules.  And the more sophisticated the module,
the more "other stuff" you need.

This combines with the fact that the customization for gnulib modules
is increasingly encoded in makefile snippets, rather than being managed
via the C preprocessor, and it means that supporting non-POSIX-based
environments is getting more difficult all the time.

I certainly would not mind having the can of worms turned into
fertilized soil (to torture a metaphor) but I suspect this will be a
LOT more effort than you expect.

Reply via email to