524984 - Electrolysis should build and run on gcc 4.2

Status: RESOLVED FIXED

(Firefox Build System :: General, defect)

Description

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Probably want to test with FC8 and maybe Ubuntu 8.04, at least.

Comment 1

[Benjamin Smedberg]

Run, yes. Build, I don't care quite so much. Do we know what's not building correctly? Is it something we can configure test to at least error out early instead of producing weird results halfway through a build, or even just not working quite right?

Comment 2

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

I think Mozilla developers are going to be annoyed if the e10s merge requires them to upgrade their linux distros.

Comment 3

[Benjamin Smedberg]

Let's figure out exactly what the problem is. Smaug was using -ggdb -feliminate-unused-debug-symbols and most of his problems went away when he stopped using those, right? It's not clear to me that we actually have a problem yet, or what its scope is.

Comment 4

[Benjamin Smedberg]

This doesn't block landing in m-c, though it may block enabling the pref by default.

Comment 5

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

(In reply to comment #4)
> This doesn't block landing in m-c, though it may block enabling the pref by
> default.

Since I don't think we want to land on m-c with maemo-fenntrolysis disabled, I think this should block.

Comment 6

[Benjamin Smedberg]

The particular issue in bug 525677 sure does, since it turns the tinderbox red! But I still don't think this bug is well-specified enough to block anything.

Comment 7

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

According to [1], our scratchbox uses gcc 4.2.1.  If that's true, then fixing bug 525677 isn't going to turn the tree green, because we should fail to link any binaries.  I can confirm that that's true for Ubuntu 8.04 / gcc 4.2.4 on x86.

[1] http://maemo.org/development/sdks/maemo_5_beta_2_sdk_installation/

Comment 8

[Benjamin Smedberg]

It built fine until .asyncDrawXULElement landed. I don't think anyone has actually tested those binaries to see if they run.

Comment 9

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Attachment: build fixes

|explicit| copy ctors have different semantics in gcc 4.2, go to the lowest common denominator

Comment 10

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Comment on attachment 410083 [details] [diff] [review]
build fixes

Thought I had these in different patches.  Anyways, the other fix was avoiding the typeinfo visibility bug, by ahem, hackefarious purposes.  I'm not really sure what the best "real" fix is ... configure.in test?

Comment 11

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

The IPDL fix was minor and pushed along with http://hg.mozilla.org/projects/electrolysis/rev/961bfa23f570

Comment 12

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

We should be golden on 4.2 now, closing.