Re: git: d22f15315b93 - main - devel/gmake43: Corrections

On Fri 29 Mar 21:38, Baptiste Daroussin wrote:
> On Fri 29 Mar 20:07, Yuri Victorovich wrote:
> > The branch main has been updated by yuri:
> > 
> > URL: https://cgit.FreeBSD.org/ports/commit/?id=d22f15315b93dcbb96df76f9a4555ff201814fb6
> > 
> > commit d22f15315b93dcbb96df76f9a4555ff201814fb6
> > Author:     Yuri Victorovich <yuri@FreeBSD.org>
> > AuthorDate: 2024-03-29 20:02:40 +0000
> > Commit:     Yuri Victorovich <yuri@FreeBSD.org>
> > CommitDate: 2024-03-29 20:02:40 +0000
> > 
> >     devel/gmake43: Corrections
> >     
> >     Reported by:    Gleb Popov <arrowd@freebsd.org>
> > ---
> >  devel/gmake43/Makefile | 5 +----
> >  1 file changed, 1 insertion(+), 4 deletions(-)
> > 
> > diff --git a/devel/gmake43/Makefile b/devel/gmake43/Makefile
> > index f012ce4c0cec..508436255675 100644
> > --- a/devel/gmake43/Makefile
> > +++ b/devel/gmake43/Makefile
> > @@ -5,10 +5,7 @@ MASTER_SITES=	GNU/make
> >  DISTNAME=	make-${PORTVERSION}
> >  PKGNAMESUFFIX=	43
> >  
> > -# note: before committing to this port, contact portmgr to arrange for an
> > -# experimental ports run.  Untested commits may be backed out at portmgr's
> > -# discretion.
> > -MAINTAINER=	tijl@FreeBSD.org
> > +MAINTAINER=	yuri@FreeBSD.org
> >  COMMENT=	GNU version of 'make' utility
> >  WWW=		https://www.gnu.org/software/make/
> 
> This is wrong, because it conflicts with the regular gmake if you want to
> actually fix the issue of performance you faced, it would have been a good idea
> to start a discussion and 1/ the gmake maintainer, 2/ ports community including
> portmgr.
> 
> Just speaking on what should be done if we are to revive gmake 4.3 first it
> needs to be done in a way which prevent collision with regular gmake: meaning
> the binary and all the installation should be in a different path and/or name
> than what the gmake provides like naming the binary gmake43.
> 
> Second the framework (Uses/gmake.mk for instance) should grow a new option to
> chose gmake43, last it should only be used on ports where the regression has
> clearly be characterized.
> 
> that is if we consider it is worth importing, which I don't think it is right
> now.
> 
> From what I could read we have a port seems to be abusing gmake and should
> probably be fixed to work properly with gmake43, doing this will benefit both
> upstream and us, so in my opinion we should not import here gmake43 neither
> follow the path I described above, but work with the only port which has an
> issue with gmake 4.4+

As I have been told by other this is not explicit enough, an explicit version of
this email is: please revert gmake43 addition we are not at a point where it is
actually required.

Best regards,
Bapt