On 2017/07/25 12:18, attila wrote: > > Stuart Henderson <s...@spacehopper.org> writes: > > > On 2017/07/24 22:33, Landry Breuil wrote: > >> myeah, i dont really have an opinion on that mess, but the autoconf goo > >> is going away someday. I see that all sub-ports were also bumped so we > >> shouldnt have update issues. No opinion on the update itself, it looks > >> fine portswise (i'm just not sure at all the gettext module is needed > >> anymore) > > > > Yep we're trying to get rid of gettext module use, I'd rather not re-add it > > as this diff does. > > Trying a build with no devel/gettext now, although it sounds like it's > certainly going to be okay... just like to make sure. > > >> - someone running this should take care of properly testing it. > > > > Presumably Sean has already done that so we are probably just down to > > ports nitpicking now :) > > While nitpicking, I prefer "attila" to "Sean" :-)
Ah sorry - noted. > As for testing, no, I'm afraid you're wrong: I was not able to do > nearly as much testing as is needed, which is why I called for testers > in my original post. The one person who tests my TB builds regularly > was seeing some bizarre behavior with regard to the toolbars: the > normal firefox personal bookmarks toolbar was appearing for him, and I > can't explain why or how. I could not get this to happen for me, but > my ability to test is limited due to my crap network connection. > > So: I really need testers for this. I don't normally like to post an > update that hasn't been tested better than this but it was simply > taking too long to get this update out, I got lapped by upstream twice > as it is and I figured there must be others out there who are > interested in testing. > > > : -# there are sometimes .orig files in the source, messes with > > update-patches > > : post-extract: > > : + # there have been .orig files in source, messes with > > update-patches > > : -find ${WRKSRC} -name '*.orig' -exec rm {} \; > > > > Probably better to set PATCHORIG, this will avoid the need for running this > > find(1) across the source tree for each build. > > I didn't know about PATCHORIG. Thanks! > > > : # relies on pulseaudio for sound and broken at runtime > > : -CONFIGURE_ARGS += --disable-webrtc > > : - > > : -CONFIGURE_ARGS += --disable-gstreamer > > : +#CONFIGURE_ARGS += --disable-webrtc > > > > It looks a bit odd to have a commented-out --disable-webrtc here, was > > that intentional? > > Much of the churn in this update's browser/Makefile was to try to get > back closer to www/firefox-esr/Makefile, to make it easier to see > changes. This bit was along those lines. I can remove it if desired, > but in general I like to stay as close to landry@'s Makefile as I can. Ah - so I'd suggest something like WANTLIB += X11-xcb xcb xcb-shm intl iconv ${LIBECXX} LIB_DEPENDS += devel/gettext (the same makes sense for firefox-esr, probably simplest to do this for tbb first and then copy it across) > >> I think you can remove the skia/gfx patches for i386, iirc they're not > >> needed in 52. To be tested. > > > > Since they're no longer present in firefox-esr I would be ok with just > > dropping them, if I run into problems in i386 bulk I can always re-add them. > > > I did remove them, using cvs rm; they look removed to me in my patch > (+++ /dev/null)... did I mess it up? You didn't - I think landry and I both misread the diff :) > I'm waiting to confirm that the build w/o devel/gettext looks good and > then I'll post another update that uses PATCHORIG. If nobody cares > about the files/configure issue maybe I'll leave it the way it was > originally: one more thing in files/ vs two new patches. If there > are feelings the other way please say so. I am happy with that.