Created attachment 359198 [details, diff] patch against gx86 Minor changes: - added EPREFIX in some places - ${D} -> ${ED}
Comment on attachment 359198 [details, diff] patch against gx86 >--- /usr/portage/eclass/texlive-common.eclass 2013-06-26 10:01:15.000000000 -0600 >+++ texlive-common.eclass 2013-09-21 12:38:25.976209928 -0600 >@@ -27,16 +27,17 @@ > # configuration > > texlive-common_handle_config_files() { >+ has "${EAPI:-0}" 0 1 2 && ! use prefix && ED="${D}" && EPREFIX= All texlive ebuilds are EAPI 3+ I think so I'd rather add a die in global scope to force EAPI 3+ but please double check texlive-common.eclass is a bit more widely used but it should be ok to bump EAPI requirement, after converting remaining ebuilds, to 3+ http://qa-reports.gentoo.org/output/eapi-per-eclass/ seems useless atm apart from this it looks ok
(In reply to Alexis Ballier from comment #1) > Comment on attachment 359198 [details, diff] [details, diff] > patch against gx86 > > >--- /usr/portage/eclass/texlive-common.eclass 2013-06-26 10:01:15.000000000 -0600 > >+++ texlive-common.eclass 2013-09-21 12:38:25.976209928 -0600 > >@@ -27,16 +27,17 @@ > > # configuration > > > > texlive-common_handle_config_files() { > >+ has "${EAPI:-0}" 0 1 2 && ! use prefix && ED="${D}" && EPREFIX= > > All texlive ebuilds are EAPI 3+ I think so I'd rather add a die in global > scope to force EAPI 3+ > but please double check > > texlive-common.eclass is a bit more widely used but it should be ok to bump > EAPI requirement, after converting remaining ebuilds, to 3+ > http://qa-reports.gentoo.org/output/eapi-per-eclass/ seems useless atm > > apart from this it looks ok <http://qa-reports.gentoo.org/output/eapi-per-eclass/texlive-common.eclass/STATS.txt> says 0 count ? Anyhow would prefer using the patch like is, EAPI<3 is deprecated in gx86, so this problem will solve itself later, once these EAPI are banned. Then the "has "${EAPI:-0}" 0 1 2 && ! use prefix" can be removed from all eclasses.
(In reply to Christoph Junghans from comment #2) > (In reply to Alexis Ballier from comment #1) > > Comment on attachment 359198 [details, diff] [details, diff] [details, diff] > > patch against gx86 > > > > >--- /usr/portage/eclass/texlive-common.eclass 2013-06-26 10:01:15.000000000 -0600 > > >+++ texlive-common.eclass 2013-09-21 12:38:25.976209928 -0600 > > >@@ -27,16 +27,17 @@ > > > # configuration > > > > > > texlive-common_handle_config_files() { > > >+ has "${EAPI:-0}" 0 1 2 && ! use prefix && ED="${D}" && EPREFIX= > > > > All texlive ebuilds are EAPI 3+ I think so I'd rather add a die in global > > scope to force EAPI 3+ > > but please double check > > > > texlive-common.eclass is a bit more widely used but it should be ok to bump > > EAPI requirement, after converting remaining ebuilds, to 3+ > > http://qa-reports.gentoo.org/output/eapi-per-eclass/ seems useless atm > > > > apart from this it looks ok > <http://qa-reports.gentoo.org/output/eapi-per-eclass/texlive-common.eclass/ > STATS.txt> says 0 count ? > > Anyhow would prefer using the patch like is, EAPI<3 is deprecated in gx86, > so this problem will solve itself later, once these EAPI are banned. Then > the "has "${EAPI:-0}" 0 1 2 && ! use prefix" can be removed from all > eclasses. yes but this is just noise I do not want to have to maintain :) IOW: please check eclass usage per EAPI and make EAPI3 mandatory for these; I don't want to have to do it later since that means it'll never be done...
(In reply to Alexis Ballier from comment #3) > yes but this is just noise I do not want to have to maintain :) That is understandable! > > IOW: please check eclass usage per EAPI and make EAPI3 mandatory for these; > I don't want to have to do it later since that means it'll never be done... Even if I would bump all ebuilds, which will take forever waiting for agreement of all the individual maintainers, we would have to wait until all EAPI<3 ebuild leave gx86, which might involve a couple of stabilization, which will make this take even longer. For this reason I won't do anything right now and leave this patch here as remember to fix this in a couple of years.
(In reply to Christoph Junghans from comment #4) > Even if I would bump all ebuilds, which will take forever waiting for > agreement of all the individual maintainers, we would have to wait until all > EAPI<3 ebuild leave gx86, which might involve a couple of stabilization, don't be so pessimistic, I'm almost sure I maintain all the consumers of these eclasses :) texlive ebuilds which would have been the main blockers are all eapi3+ afaik, this leaves only a couple of minor packages for which the eapi bump can likely be done in place
Maybe it isn't that bad: - app-text/jadetex - no EAPI (=0), stable - dev-tex/xmltex - no EAPI (=0), stable - app-text/active-dvi - EAPI=2, ~arch
(In reply to Christoph Junghans from comment #6) > Maybe it isn't that bad: > - app-text/jadetex - no EAPI (=0), stable bump all of them to EAPI5; this one should be as simple as setting EAPI and removing now useless || die at a quick look; attach a diff if unsure > - dev-tex/xmltex - no EAPI (=0), stable ditto > - app-text/active-dvi - EAPI=2, ~arch ditto but also add a := dep on dev-lang/ocaml and dev-ml/camlimages; nfc why I haven't committed this yet...
(In reply to Alexis Ballier from comment #7) ah and maybe you'll want to add EPREFIX/ED stuff to these ebuilds too
(In reply to Alexis Ballier from comment #8) > (In reply to Alexis Ballier from comment #7) > > ah and maybe you'll want to add EPREFIX/ED stuff to these ebuilds too This is only needed for app-text/active-dvi, the other two packages already have prefix keywords.
Created attachment 368838 [details, diff] patch against gx86 All depending packages updated to EAPI>=3.
ping @aballier!
(In reply to Christoph Junghans from comment #10) > Created attachment 368838 [details, diff] [details, diff] > patch against gx86 > > All depending packages updated to EAPI>=3. oops; i had missed that one patch lgtm, feel free to commit and next time please attach unified diffs, those are easier to read
+ 18 Mar 2014; Christoph Junghans <ottxor@gentoo.org> texlive-common.eclass: + added prefix support (bug #485608) +
