Re: Overhauling gcc.mk

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

* On 2023-10-19 at 07:31 BST, nia wrote:

> Looks good to me, thanks for working on this.
>
> Is removing the check for specific NetBSD versions "ok"?
> Will the appropriate -libs package be automatically depended
> on if a newer gcc package is pulled in to compile something
> and USE_PKGSRC_GCC isn't set?

Yep.  What I'd also like to do after this is to simply make 
USE_PKGSRC_GCC_RUNTIME=yes the default.  It's highly unlikely that 
anyone would not want the behaviour, but if for whatever reason they do 
they can turn it off, and this further simplifies things.  I'll also 
tidy up some of the logic in the -libs packages themselves to avoid the 
whole PKGREVISION of gcc-libs > gcc thing.

I've also removed USE_GCC_RUNTIME in a recent push.

I've tried to explain the USE_NATIVE_GCC / USE_PKGSRC_GCC logic in some 
block comments to make it hopefully a lot clearer what they actually do 
in reality.  Leaving USE_PKGSRC_GCC unset doesn't stop it being pulled 
in if necessary.  It's only when USE_PKGSRC_GCC=yes that the situation 
changes and pkgsrc GCC will be used regardless of whether the native GCC 
matches GCC_REQD.  This is where the user may need GCC_BOOTSTRAP_PKGS to 
exclude packages in the pkgsrc GCC bootstrap path.

I'm happy to reword the block comments to make it even clearer, so let 
me know if any of this is still unclear in any way.

I've been testing this in some bulk builds and it currently seems to be 
working well, but will also do some static analysis of a bunch of 
variables before and after on NetBSD etc.

Cheers,

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