On Thu, Apr 09, 2020 at 01:23:24PM +0200, Tim Tassonis via blfs-dev wrote: > Hi all > > I noticed that I do get quite different build sizes for thunderbird than > other maintainers, so I thought I will give a few infos about how I build > thunderbird. Maybe this helps to find out what's going on here. > Hi Tim,
thanks for providing this information. > > > du -hs /lgl-bld/thunderbird-68.7.0/ > > > reports > > 4.2G /lgl-bld/thunderbird-68.7.0/ > > > So, as I'm doing this with all other packages and never get a significant > difference to what's in the book with other packages, I'm pretty sure that > my counting isn't the problem. > > In my mozconfig, I think I'm doing pretty standard stuff, too. These are my > active ac_add_options: Just for the record, on firefox my own builds use harder CFLAGS etc, potentially different optimizations, and (while 68-esr lasts) the patch for system graphite2 and harfbuzz. But when I measure for the book I match the book's settings (without the optional patch) and without my own CFLAGS etc. So, I don't think that any differences in the mozconfig you are using will be the cause of this enormous difference in size. > > ac_add_options --enable-startup-notification > ac_add_options --disable-pulseaudio > ac_add_options --enable-calendar > ac_add_options --enable-system-sqlite > ac_add_options --with-system-libevent > ac_add_options --with-system-nspr > ac_add_options --with-system-nss > ac_add_options --with-system-icu > ac_add_options --prefix=/opt/thunderbird > ac_add_options --enable-application=comm/mail > ac_add_options --disable-crashreporter > ac_add_options --disable-updater > ac_add_options --disable-debug > ac_add_options --disable-debug-symbols > ac_add_options --disable-tests > ac_add_options --enable-optimize=-O2 > ac_add_options --enable-strip > ac_add_options --enable-install-strip > ac_add_options --enable-official-branding > ac_add_options --enable-system-ffi > ac_add_options --enable-system-pixman > ac_add_options --with-system-bz2 > ac_add_options --with-system-jpeg > ac_add_options --with-system-png > ac_add_options --with-system-zlib > > > > So, what's left maybe is my toolchain, which might be a bit outdated: > > Name : rustc > Version : 1.42.0-1 > I had thought you must be using an older version (back to 1.35 is probably still usable with 68). > Name : gcc > Version : 7.3.0-1 > > Name : clang > Version : 5.0.1-1 > I'm shocked that your versions are so old. > I suppose I'm building with gcc and not clang, might that make such a huge > difference in buildsize? > In firefox there were past variations, issues with gold, etc and eventually I got to a point where building with the current gcc saved a lot of space in the build. If I used thunderbird other than for testing if we can upgrade rust, I would have pressed to try moving it to gcc in the belief that space would be saved. But I don't think that either gcc-7 or clang-5 are viable choices nowadays. Re clang I was going to remove some old coments from my firefox upgrade script the other day, but decided to keep them. In the section about upgrading rustc if it was too old (for old LFS releases I try to keep usable) I note: clang-5 was too old to build some past version of rust with sysllvm. clang-6 and -7 were unreliable with firefox (particularly on my 4-core ryzen, they tended to fall back to 1 core) But all of that is now, for me, ancient history - my oldest system is based on LFS-8.3. And from that I have scripts to build rustc either with sysllvm or with its own shipped llvm. For your 1.42.0 I assume you are using the shipped llvm. Overall, I think your main toolchains are the problem. > > Bye > Tim ĸen -- The beauty of reading a page of de Selby is that it leads one inescapably to the conclusion that one is not, of all nincompoops, the greatest. -- du Garbandier -- http://lists.linuxfromscratch.org/listinfo/blfs-dev FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page