798789 - Enable gcc -Werror=conversion-null for gcc >= 4.5 for comm-central (port bug 778980)

Status: RESOLVED FIXED

(MailNews Core :: Build Config, defect)

Description

[:aceman]

Apply this to comm-central if it is useful.

+++ This bug was initially created as a clone of Bug #778980 +++

To make gcc's NULL/nullptr warnings fatal like clang's nullptr errors, we may want to add -Werror=conversion-null to Linux's CFLAGS/CXXFLAGS for gcc >= 4.5. (The -Wconversion-null flag was new in gcc 4.5.)

If a developer overlooks these gcc warnings, then they may not realize their code change will break the OS X's clang build until they push to the tryserver. This is a waste of developer time and tryserver resources.

Comment 1

[:aceman]

Attachment: patch

Here is the patch. Archaeopteryx, please try it on a try server (linux only is enough). I did not get any failures, but I compile with many components disabled.

Comment 2

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

Thunderbird-Try run: https://tbpl.mozilla.org/?tree=Thunderbird-Try&rev=408d5cf11a05
Builds are green

Comment 3

[:aceman]

Comment on attachment 668800 [details] [diff] [review]
patch

The build try server was done for Thunderbird only. Can anybody try Seamonkey?

Comment 4

[Mark Banner (:standard8)]

SeaMonkey Try run: https://tbpl.mozilla.org/?tree=Thunderbird-Try&rev=752bf0e94218

Comment 5

[:aceman]

Thanks for the build.

There is a configure error:
configure: error: Unrecognized extension provided to --enable-extensions: venkman.

That does not seem caused by this patch.

Comment 6

[Mark Banner (:standard8)]

Comment on attachment 668800 [details] [diff] [review]
patch

The first SeaMonkey run I got wrong, but the second should work:

https://tbpl.mozilla.org/?tree=Thunderbird-Try&rev=1e27f1fd450a

Comment 7

[Mark Banner (:standard8)]

(and that's r+ assuming that run passes).

Comment 8

[:aceman]

Now an "Internal server error" and another run with "abort: connection ended unexpectedly" ???

Comment 9

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

If you this happens during |hg pull -u|:
That happens often. The more changesets, the more likely. Just try again.

Comment 10

[Mark Banner (:standard8)]

If they are blue, they'll automatically be re-run. The builds finished off red in stage-package, but that's because SM is expecting to package the extensions that I disabled (rather than having them pulled). So the actual compilations did pass. That's about the best we can do with SM on try server.

Comment 11

[:aceman]

OK, thanks. So I take it that the build works.

If any breakage appears later please point me to it.

Comment 12

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/comm-central/rev/ccb28f98f40a

Comment 13

[:aceman]

Ryan, any reason you did not close this bug after check-in?

Comment 14

[Ryan VanderMeulen [:RyanVM]]

Nope, just forgot. Thanks.