Re: emacs packages fail

[david.sainty%dtsp.co.nz@localhost]

david.sainty writes:

> Aleksey Cheusov writes:
> 
> >  >> But EMACS_VERSION_MAJOR is empty.
> > 
> >  >>    0 ! emacs-w3m>~/pkg.temp/bin/bmake show-var 
> > VARNAME=EMACS_VERSION_MAJOR
> > 
> > 
> >  >>    0 ! emacs-w3m>
> > 
> > > I've already reported this bug about year ago but it is still not fixed.
> > > http://www.netbsd.org/cgi-bin/query-pr-single.pl?number=37146
> > 
> > > Result of 'make show-var VARNAME=DEPENDS' should not depend
> > > on whether emacs is installed or not.
> > 
> > It's not clear for me what does this silence mean. About year ago I've
> > reported a bug. Now I provide a solution, patch that helps to detect
> > emacs version without running 'pkg_info -e <emacs>' by introducing
> > <emacses>/Makefile.ver files and changing mk/emacs.mk appropriately.
> > 
> > I'd like to see PR/37146 fixed and my solution is easy.  Patch is also
> > trivial, several Makefile.ver appeared and mk/emacs.mk changed
> > accordingly.
> > 
> > So, what's the problem? :-( Is patch bad? Tell why.
> 
> It probably means most people aren't being bitten by this.  But I am
> :)  Your solution looks pretty reasonable to me, it's pretty much how I
> was thinking of structuring a fix to it last time I was mulling it.
> 
> I'll test it and commit it [before the freeze or after the freeze, but
> not during :)] if there's no dissenting comments in the meantime...

Prior art says that "Makefile.version" is preferred over
"Makefile.ver".  I'm not sure that a variable with a leading
underscore (_EMACS_VERSION) belongs in package files, but there seems
to be a bit of prior art for that...