On Mon, Dec 11, 2017 at 09:33:28PM +0100, Landry Breuil wrote:
> On Mon, Dec 11, 2017 at 07:45:06PM +0000, Bryan C. Everly wrote:
> > Hello ports@
> >
> > Taking a second attempt at this. I have updated per Rafael Sadowski
> > to remove REVISION and I have placed the sample luminaDesktop.conf
> > file in the correct place (versus the FreeBSD place) per Landry
> > Breuil.
> >
> > As mentioned before, attached is a diff that gets us current with the
> > latest release of the Lumina Desktop Environment (a BSD-licensed
> > desktop environment). At this point, since we have updated our
> > version of qt5, we now have all of the same binaries as upstream.
> >
> > I'd appreciate some further eyes on the diff and then, if it looks
> > good, some OK's and a commit.
>
> It *looks* good minor two nits:
>
> > -@unexec-delete rm -f %D/share/icons/material-design-dark/icon-theme.cache
> > -@unexec-delete rm -f %D/share/icons/material-design-light/icon-theme.cache
>
> Here you lose those two lines
>
> > -share/examples/lumina/luminaDesktop.conf
> > -@sample ${SYSCONFDIR}/luminaDesktop.conf
>
> ...
>
> > share/icons/material-design-dark/
> > +share/examples/lumina/luminaDesktop.conf
>
> And here you lost the @sample (and the share/examples line was
> reordered/put at the wrong place)
>
> Never. Ever. Blindly trust make update-plist. It *always* need manual
> checking/extra tweaks, especially wrt @xxx annotations.
Oh, and now that i thought a bit more about your diff (that i haven't
actually *tested* yet), i don't see any change in the build
config/patches, so how have you 'placed the sample luminaDesktop.conf
file in the correct place' ? If you manually edit PLIST, that's not how
it works...
make fake will run 'make install' in the work dir, pointing at a
temporary fake hierarchy, and make update-plist will list files in the
given hierarchy, and update PLIST with files it found there. So if you
want to fix the install location, you have to rewind a step before, and
either fix the arguments passed to the lumina build system to make sure
it installs this config file at the right place (not hardcoding paths
inside the code to the 'wrong' location, since the actual location at
runtime will be the @sample'd one) or patch the makefiles so that it
uses the 'runtime' location inside the code (i wouldnt be surprised it's
hardcoded somewhere in the code) and adapt only the install path during
fake.
dammit, sometimes, ports is hard. Maybe re-read the ports faq and
pkg_create ?
Oh and of course maybe it's actually fixed the right way already, and
i'm talking out of my ***.
