734080 - Port configure.in changes from Bug 700910 to Thunderbird

Status: RESOLVED FIXED

(MailNews Core :: Build Config, defect)

Description

[:Irving Reid (No longer working on Firefox)]

Core recently added support for the C++11 'delete' keyword on method declarations; this requires a new warning control flag on clang.

Comment 1

[:Irving Reid (No longer working on Firefox)]

Also pulling in bug 666135, which defines the CLANG_CXX compiler detection and takes out some of the other bothersome warnings

Comment 2

[:Irving Reid (No longer working on Firefox)]

Attachment: configure changes to reduce spurious warnings under clang

Clang detection and warning suppression ported from mozilla-central, and drive-by cleanup of DOS line endings in an unrelated file. Try build on its way.

Comment 3

[Justin Wood (:Callek)]

Comment on attachment 604153 [details] [diff] [review]
configure changes to reduce spurious warnings under clang

review @ mark on irvings behalf.

Comment 4

[:Irving Reid (No longer working on Firefox)]

TBPL shows one of the try builders not completing, but all the others look fine: http://build.mozillamessaging.com/tinderboxpushlog/?tree=ThunderbirdTry&rev=31e66a799604

Comment 5

[:Irving Reid (No longer working on Firefox)]

Comment on attachment 604153 [details] [diff] [review]
configure changes to reduce spurious warnings under clang

Review of attachment 604153 [details] [diff] [review]:
-----------------------------------------------------------------

A bunch more configure.in changes just dropped in bug 711895, I'll respin the patch and run another try build.

Comment 6

[:Irving Reid (No longer working on Firefox)]

This can land before bug 737677 but some of the warnings won't go quiet until then

Comment 7

[:Irving Reid (No longer working on Firefox)]

Attachment: Port configure changes from M-C for better compiler warning control

M-C has been doing a bunch of work on their compiler warnings, not all of them clang-specific. This patch pulls in the most recent batch.

Switching reviewer to bienvenu to give standard8 a break.

Comment 8

[David :Bienvenu]

Comment on attachment 607987 [details] [diff] [review]
Port configure changes from M-C for better compiler warning control

what's going on with db/makefiles.sh? line ending changes? should that not be part of this diff?

Comment 9

[:Irving Reid (No longer working on Firefox)]

Yes, db/makefiles.sh is a drive-by fix of DOS line endings in an unrelated file. If you want to leave that out of this patch that's fine.

Comment 10

[David :Bienvenu]

Comment on attachment 607987 [details] [diff] [review]
Port configure changes from M-C for better compiler warning control

builds fine on windows - db/makefiles.sh change is ok, just checking.

Comment 11

[Nicholas Nethercote [inactive]]

Please note that bug 711895 caused some i10n build bustage and may be backed out again :(

Comment 12

[:Irving Reid (No longer working on Firefox)]

Comment on attachment 607987 [details] [diff] [review]
Port configure changes from M-C for better compiler warning control

Patch has suffered bit-rot, update coming...

Comment 13

[:Irving Reid (No longer working on Firefox)]

Attachment: Port configure changes from M-C for better compiler warning control, updated

No substantive changes, just bit-rot from other configure patches that landed in between. Try is busy (and a bit flaky), but I can queue this up if you'd like to see a try build before landing.

Comment 14

[:Irving Reid (No longer working on Firefox)]

Try build at http://build.mozillamessaging.com/tinderboxpushlog/?tree=ThunderbirdTry&rev=e2d1a8c405b5 looks green enough to me, let's push this and see what comes out the other side.

Comment 15

[Ryan VanderMeulen [:RyanVM]]

http://hg.mozilla.org/comm-central/rev/014d81b1029f

Comment 16

[Justin Wood (:Callek)]

BACK THIS OUT!

p.s. given the issues mentioned in the other bug, I'm surprised bienvenu was ok with reviewing this, since he's not technically a build-system peer for c-c.

We have had broken l10n builds since this landed on trunk (on linux). 

Please get review from a c-c actual build system peer (or an m-c build system peer if they are willing to review this) before relanding.

Comment 17

[Justin Wood (:Callek)]

pps I'm surprised at "us" for not catching that this did break repacks, especially given c#11, and sad that no-one happened to tell you that try is not good enough to catch l10n repack bustage from the change.

Comment 18

[:Irving Reid (No longer working on Firefox)]

Attachment: Additional changes (on top of previously committed patch) to fix l10n repack

This patch should be applied on top of https://bugzilla.mozilla.org/attachment.cgi?id=612185 and contains additional changes from the updated version of bug 711895, to support l10n repack builds using older versions of GCC. Try build in progress at http://build.mozillamessaging.com/tinderboxpushlog/?tree=ThunderbirdTry&rev=35b9036dc1d1

Comment 19

[Ryan VanderMeulen [:RyanVM]]

Revision 014d81b1029f backed out. I'll re-land both together.
http://hg.mozilla.org/comm-central/rev/f62eaa1c5b0d

Comment 20

[:Irving Reid (No longer working on Firefox)]

First try build seems to have suffered from some mozilla-central bustage; second try came through clean aside from a couple of random-looking issues: http://build.mozillamessaging.com/tinderboxpushlog/?tree=ThunderbirdTry&rev=73154769b254

Comment 21

[Justin Wood (:Callek)]

Comment on attachment 614489 [details] [diff] [review]
Additional changes (on top of previously committed patch) to fix l10n repack

Review of attachment 614489 [details] [diff] [review]:
-----------------------------------------------------------------

Looks good sorry for the delay

Comment 22

[Justin Wood (:Callek)]

Need to reland both patches on this bug...

Comment 23

[Ryan VanderMeulen [:RyanVM]]

http://hg.mozilla.org/comm-central/rev/247aa759c1ce
http://hg.mozilla.org/comm-central/rev/134579274546