Tracking bug for MozillaBuild 1.9. So far: * Various improvements for newer MSVC and Windows SDK versions * Updated Mercurial * Mozmake (GNU make 4.0 + custom patches) * Rearrange directory structure to be less cluttered * Ensure that all in-tree sources are present and up-to-date
Update Mercurial to version 2.8.0 (I expect to get 2.8.1 in before this is officially released). https://hg.mozilla.org/mozilla-build/rev/b451d0b26ccb
Fix Mercurial.rc.patch (bitrotted by 2.8.0). http://hg.mozilla.org/mozilla-build/rev/df2cb0eea59a
Here's a test build. I'll post a note to dev-platform too. http://people.mozilla.org/~rvandermeulen/MozillaBuildSetup1.9.0pre.exe SHA1: 4F227BE966E8AE73702C6E211B15064ED15D44C9
Assignee: nobody → ryanvm
Is it possible that the MozillaBuild use msys2, since it's up-to-date? I can get successful builds with it. https://sourceforge.net/projects/msys2
Update Mercurial to version 2.8.2: http://hg.mozilla.org/mozilla-build/rev/3b70c5cb46b7 (In reply to zhoubcfan from comment #4) > Is it possible that the MozillaBuild use msys2, since it's up-to-date? I can > get successful builds with it. > > https://sourceforge.net/projects/msys2 I'd like to update MSYS at some point, but probably not for this release.
(In reply to Ryan VanderMeulen [:RyanVM UTC-5] from comment #5) > Update Mercurial to version 2.8.2: > http://hg.mozilla.org/mozilla-build/rev/3b70c5cb46b7 Why did the TBPL repo get an empty commit for this too? https://hg.mozilla.org/webtools/tbpl/rev/69a65cdb3903 Was it your import/export script? If so, please could you give qimportbz/bzexport a go? :-)
Because I was an idiot and working in the wrong repo at the time.
Update Mercurial.rc.patch again. https://hg.mozilla.org/mozilla-build/rev/a945d321aefa
I was hoping to update python to version 2.7.6, but mach doesn't like it.
Ryan, can you update autoconf from 2.56 to 2.6x? Also, do you know anyone who knows how to build the MSYS bash? The MSYS bash does not extract a process's exit status correctly if the process passes a negative number to exit(). This bug affects the testing of NSS on Windows because NSS uses shell scripts to run the tests. I think this bug is not hard to track down, but I don't know how to build the MSYS bash from source code. Thanks.
(In reply to Wan-Teh Chang from comment #10) > Ryan, can you update autoconf from 2.56 to 2.6x? Unfortunately, our ancient version of MSYS (1.0.11) limits what packages we can upgrade to at this point. AFAICT, autoconf 2.61 is the most recent version that has a compatible build. Would that work? FWIW, I have longer-term hopes of switching us over to MSYS2, which gives us a much better upgrade story. But that's not on the 1.x horizon :) > Also, do you know anyone who knows how to build the MSYS bash? > The MSYS bash does not extract a process's exit status correctly > if the process passes a negative number to exit(). This bug > affects the testing of NSS on Windows because NSS uses shell > scripts to run the tests. I think this bug is not hard to track > down, but I don't know how to build the MSYS bash from source > code. Thanks. I don't know, sorry.
Ryan: An autoconf 2.5x bug that I need to work around was fixed in 2.60, so 2.61 will allow me to remove my workaround. On the other hand, I know 2.60 seems to contain some major rewrites, so I am not sure if 2.61 is stable enough. Based on the release dates of 2.60, 2.61, and 2.62, I think 2.61 looks like a stabilization release of 2.60, and 2.62 looks like a feature release. So updating to 2.61 should be fine. Thanks.
If we build autoconf-2.61 for mozilla-build, does it work fine? autoconf-2.13 wasn't included in msys, we built autoconf-2.13 by packageit.sh.
Makoto: I'm sorry that I don't understand your question. MozillaBuild 1.8.0 includes both "autoconf-2.13" and "autoconf", which is autoconf 2.56. Yesterday I figured out how to port nsprpub/configure.in to autoconf 2.56. I didn't try to port the other configure.in files in the Mozilla source tree. So MozillaBuild will still need to include "autoconf-2.13". I am only requesting that "autoconf" be upgraded to autoconf 2.61. Does this answer your question?
This is offtopic for this bug but since you're not pointing at a bug about that switch to autoconf 2.5+ for nspr, I'll say it here: please don't. There is a lot of duplication between mozilla configure.in and nspr configure.in, most notably for android. Switching nspr configure to autoconf 2.5 increases the chances of copy/paste utterly failing.
Update Mercurial to version 2.9.1: http://hg.mozilla.org/mozilla-build/rev/644bb0b16cf9
I'm going to use this for a couple days and shoot for a final release by the end of the week.
Windows 603 please not awailable and i can compile to 602 only ? I use MozillaBuildSetup1.9.0pre.exe on Windows 8.1 64bit, SDK 8.1, msvc2013, start-shell-msvc2013-x64.bat, and alwais have this error: configure: error: Invalid value for --with-windows-version (603) ------ config.log ------ configure:3341: checking whether cl accepts -g configure:3382: checking for c++ configure:3414: checking whether the C++ compiler (cl ) works configure:3430: cl -o conftest conftest.C 1>&5 Microsoft (R) C/C++ Optimizing Compiler Version 18.00.21005.1 for x64 Copyright (C) Microsoft Corporation. All rights reserved. cl : Command line warning D9035 : option 'o' has been deprecated and will be removed in a future release conftest.C
(In reply to Sladky Vladimir from comment #18) > Windows 603 please not awailable and i can compile to 602 only ? > > I use MozillaBuildSetup1.9.0pre.exe on Windows 8.1 64bit, SDK 8.1, msvc2013, > start-shell-msvc2013-x64.bat, and alwais have this error: > > configure: error: Invalid value for --with-windows-version (603) > ------ config.log ------ > configure:3341: checking whether cl accepts -g > configure:3382: checking for c++ > configure:3414: checking whether the C++ compiler (cl ) works > configure:3430: cl -o conftest conftest.C 1>&5 > Microsoft (R) C/C++ Optimizing Compiler Version 18.00.21005.1 for x64 > Copyright (C) Microsoft Corporation. All rights reserved. > > cl : Command line warning D9035 : option 'o' has been deprecated and will be > removed in a future release > conftest.C It isn't Mozilla-build issue. Actually we doesn't support --with-windows-version=603 yet. So please file new bug for Core - build config issue
Also, mozilla-build 1.9 can detect SDK 8.1 and can use it.
Here's an rc build of 1.9. I hope to release this as a final build this week. http://people.mozilla.org/~rvandermeulen/MozillaBuildSetup1.9.0rc.exe Changelog since 1.9: http://hg.mozilla.org/mozilla-build/pushloghtml?fromchange=8be93f0b6ef9&tochange=cfd4e36a9762
(In reply to Ryan VanderMeulen [:RyanVM UTC-4] from comment #21) > Changelog since 1.9: And of course, that should have said 1.8.
Thanks to a last-second crash-landing from bug 980697, I've updated the release tag to pick up that change. http://hg.mozilla.org/mozilla-build/rev/190992f68fc9 The new installer has been uploaded. The revised version has a sha1 of aec35019e2bfa51c927d1923a520773bee4b5feb for those who want to confirm that they have the latest. Sorry for the hassle.
Status: NEW → RESOLVED
Last Resolved: 4 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.