335949 - typo in fix for 319012

Status: RESOLVED FIXED

(Firefox Build System :: General, defect)

Description

[Walter Meinl]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9a1) Gecko/20060409 Firefox/3.0a1
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9a1) Gecko/20060409 Firefox/3.0a1

due to a typo in patch for bug 319012 see attachment https://bugzilla.mozilla.org/attachment.cgi?id=220144
I see following during configure process

checking for visibility(hidden) attribute... yes
checking for visibility(default) attribute... yes
checking for visibility pragma support... yes
checking For gcc visibility bug with class-level attributes (GCC bug 26905)... i686-pc-linux-gnu-g++: -z: linker input file u
nused because linking not done
i686-pc-linux-gnu-g++: defs: linker input file unused because linking not done
yes
checking For x86_64 gcc visibility bug with builtins (GCC bug 20297)... i686-pc-linux-gnu-gcc: -z: linker input file unused because linking not done
i686-pc-linux-gnu-gcc: defs: linker input file unused because linking not done
grep: @PLT: invalid context length argument

./configure: line 7915: test: ==: unary operator expected
no

This is caused by a typo in the patch
it should read configure.in

2666 if test `grep -c "@PLT" conftest.S` == 0; then

instead of 
2666 if test `grep -C "@PLT" conftest.S` == 0; then

Patch to come in a few minutes

then the respective output of configure looks
checking for visibility(hidden) attribute... yes
checking for visibility(default) attribute... yes
checking for visibility pragma support... yes
checking For gcc visibility bug with class-level attributes (GCC bug 26905)... i686-pc-linux-gnu-g++: -z: linker input file unused because linking not done
i686-pc-linux-gnu-g++: defs: linker input file unused because linking not done
yes
checking For x86_64 gcc visibility bug with builtins (GCC bug 20297)... i686-pc-linux-gnu-gcc: -z: linker input file unused because linking not done
i686-pc-linux-gnu-gcc: defs: linker input file unused because linking not done
no

as you can see no more complaints from grep, however the linker complaint remain (no clue if they are important, probably not)



Reproducible: Always

Comment 1

[Walter Meinl]

Attachment: grep -c

Comment 2

[Benjamin Smedberg]

Comment on attachment 220256 [details] [diff] [review]
grep -c 

You should always request review on patches, or they're likely to be forgotten.

Comment 3

[Walter Meinl]

ok, didn't know that I'm able to, next time

Comment 4

[Walter Meinl]

Benjamin, I could convince the Gentoo devs to use your patches for https://bugzilla.mozilla.org/show_bug.cgi?id=334866 and https://bugzilla.mozilla.org/show_bug.cgi?id=319012 (typo fixed accordingly) for ff/tb 1.5.0.2.
Are there any plans to put those fixes also on the branches (1.8.0.x would need howevar also the checkin for https://bugzilla.mozilla.org/show_bug.cgi?id=307168). Asking here, because all those bugs are marked fixed. In addition, as a newbie would't know whom to ask for approval, sorry for bothering you

Comment 5

[Benjamin Smedberg]

Comment on attachment 220256 [details] [diff] [review]
grep -c 

Duh.

Comment 6

[Benjamin Smedberg]

Fixed on trunk.

getting this on the 1.8.0 branch is pretty unlikely based on the changes to nscore and NSPR that it requires (it changed the HAVE_VISIBILITY_ATTRIBUTE/PRAGMA preprocessor defines). The workaround on the 1.8.0 branch is to add ac_cv_have_visibility_pragma=no in your mozconfig.

I need to look at the state of the 1.8 branch, it may be possible to get these fixes there (and it would be a good idea if we ever find a GCC that doesn't have these bugs).

Comment 7

[Zibi Braniecki [:zbraniecki][:gandalf]]

Benjamin: any update on getting this for 1.8?

Comment 8

[Benjamin Smedberg]

The original patch didn't hit 1.8, did it? There are complicated dependencies to resolve.

Comment 9

[Walter Meinl]

(In reply to comment #8)
> The original patch didn't hit 1.8, did it? 

Both typo corrected patches still apply cleanly as of 2006-05-30 against configure.in of MOZILLA_1_8_BRANCH just checked out. Last home made build was from May 18th. I apply those patches regularly and don't see any problems so far.