For example, http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey/1254452103.1254455054.27078.gz . Looking into it.

Changing product/component to make the right people see this. Still not sure where the blame lies.

Not tested on 1.9.1, but should work there too. Windows may need more work, but I don't have a windows build machine at the moment.

Thanks for the start Chris, I'll be taking more of a look at this in 1/2 hour or so.

This changes the if for MOZILLA_1_9_1_BRANCH to also be for 1.9.2. I've also copy-pasted the chunk that sets ac_cv_attribute_always_inline etc. As this is based on Chris' original patch, r+a=Standard8 and I'll push this now and we'll see how it affects the tree.

Changeset http://hg.mozilla.org/comm-central/rev/86f5531e2c87 has done the trick for TB Linux check at least. I'll check on Windows and Mac later.

Linux bloat seems to have crashed in its trace malloc step. I'm letting it cycle to see if it is intermittent or not. Windows builds have link error but I think I know what that is, will fix later.

I just ported some more of the configure.in changes for windows: http://hg.mozilla.org/comm-central/rev/b66d6e147cfa

Thanks Mark! Apologies again.

(In reply to comment #8) > Thanks Mark! Apologies again. No apologies necessary, we expect things changing in mozilla-central to break us from time to time.

If I didn't screw this (hg revert -r d4529d3c2429 configure.in) up, it should back out all the fixes, and green up Linux (which is the only platform I don't think is currently burning from something else besides this).

Bug 441324 doesn't seem to be going anywhere at the moment. Changing deps so we'll get notified if bug 441324 changes state, and unassigned myself as there's nothing to do here for now.

This might at least save you some time during your morning "wake up to flames." Pushed to the try server, just in case that finishes in time to tell you anything about how well I did.

Ima go out on a limb and say we want to actually *package* libmozalloc.dylib, not just build it and leave it on the floor.

Windows hasn't cycled yet, but this is looking pretty much fixed now: http://hg.mozilla.org/comm-central/rev/e1e944c8ad9e

just one [small] addition.

Comment on attachment 404389 [details] [diff] [review] Backout [Checkin: Comment 16] http://hg.mozilla.org/comm-central/rev/b41b3d8a406f

Comment on attachment 430794 [details] [diff] [review] finish-port (config.mk change) [checkin c#22] It doesn't look like c-c needs this, does it? http://mxr.mozilla.org/comm-central/search?string=MOZALLOC_LIB&case=on

(In reply to comment #14) > http://hg.mozilla.org/comm-central/rev/e1e944c8ad9e + http://hg.mozilla.org/comm-central/rev/00fe3f19f1d5 Drop unnecessary 1.9.2 ifdef in suite code added by bug 520114

Comment on attachment 430794 [details] [diff] [review] finish-port (config.mk change) [checkin c#22] I'm not completely sure if we'll ever need it, but it does no harm either.

(In reply to comment #20) > I'm not completely sure if we'll ever need it, but it does no harm either. It "harms" because I/we're in the process of removing what we don't "absolutely" needs :-/

Comment on attachment 430794 [details] [diff] [review] finish-port (config.mk change) [checkin c#22] http://hg.mozilla.org/comm-central/pushloghtml?fromchange=386d5b686699&tochange=a803df412b50

Comment on attachment 430850 [details] [diff] [review] (Fv1) Cleanup after too much or wrong port, Remove useless (legacy) MOZ_MEMORY_* r+ on the non Mozilla_Memory related removals. If you insist on those in this same patch we need Mark's sr here. If you want to do the Mozilla_Memory in a different patch, feel free to port my r+ to that, and use a new bug and sr? him.

Comment on attachment 430850 [details] [diff] [review] (Fv1) Cleanup after too much or wrong port, Remove useless (legacy) MOZ_MEMORY_* >+ MOZ_JS_LIBS='-L$(LIBXUL_DIST)/bin -lmozjs' > if test "$MOZILLA_1_9_2_BRANCH" = "1"; then >- MOZ_JS_LIBS='-L$(LIBXUL_DIST)/bin -lmozjs' >- DYNAMIC_XPCOM_LIBS='-L$(LIBXUL_DIST)/bin -lxpcom -lxpcom_core ' >+ DYNAMIC_XPCOM_LIBS='-L$(LIBXUL_DIST)/bin -lxpcom -lxpcom_core' >+else >+ DYNAMIC_XPCOM_LIBS='-L$(LIBXUL_DIST)/bin -lxpcom -lxpcom_core -lmozalloc' >+fi etc... When I reviewed this, I probably didn't comment, but I did consciously allow it in ONE block/ifdef because it was far simpler than multiple. Therefore please drop this hunk. >-MALLOC_H= >-AC_CHECK_HEADER(malloc.h, [MALLOC_H=malloc.h]) >-if test "$MALLOC_H" = ""; then >- AC_CHECK_HEADER(malloc/malloc.h, [MALLOC_H=malloc/malloc.h]) >- if test "$MALLOC_H" = ""; then >- AC_CHECK_HEADER(sys/malloc.h, [MALLOC_H=sys/malloc.h]) >- fi >-fi >-if test "$MALLOC_H" != ""; then >- AC_DEFINE_UNQUOTED(MALLOC_H, <$MALLOC_H>) >-fi I don't see why you're removing this and not the rest. There again, I'm not a wizard on this foo, but an explanation would be useful.