Closed Bug 214051 Opened 21 years ago Closed 21 years ago

configure thinks new gcc 3.3.1 has -pedantic long long bug

Categories

(SeaMonkey :: Build Config, defect)

x86
Linux
defect
Not set
major

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: calum.mackay, Assigned: mozbugs-build)

References

Details

(Keywords: fixed1.4.1)

Attachments

(1 file)

I've upgraded my compiler to the latest Debian unstable:

gcc version 3.3.1 20030722 (Debian prerelease)

which I believe is the official 3.3.1 release candidate 1. I previously had an
earlier build of gcc 3.3.1 (20030626) which configured (and built) mozilla fine.

This new build fails to configure mozilla:

checking whether C++ compiler has -pedantic long long bug... yes
configure: error: Your compiler appears to have a known bug where long long is m
iscompiled when using -pedantic.  Reconfigure using --disable-pedantic.

[This from a checkout today, 20030727; note however that I saw the error from my
first build with the upgraded compiler]

Using the noted workdaround does allow configure to complete.

My build options were:

                                --enable-crypto \
                                --enable-default-toolkit=gtk2 \
                                --enable-xft \
                                --enable-strip \
                                --enable-strip-libs \
                                --disable-tests \
                                --disable-debug \
                                --x-includes=/usr/local/X11R6/include \
                                --x-libraries=/usr/local/X11R6/lib \
                                --enable-optimize="-O2 -march=i686" \
                                --enable-extensions=spellcheck \
                                --enable-calendar

This could be a gcc bug, of course, but it still seemed sensible to bring it here.
Why file a bug on Mozilla if gcc reintroduced an old bug?  Is the test wrong?
Status: NEW → RESOLVED
Closed: 21 years ago
Resolution: --- → INVALID
That's why I filed the bug: I don't know that gcc re-introduced an old bug, nor
do I know whether the test is wrong. I thought it might prove useful for other
gcc 3.3.1 users.
I've logged gnu c++ bug 11720 to track this:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11720

no further updates here.
I'm reopening this bug.

My c++ bug (http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11720) has been closed
as invalid.

The test under question boils down to this:

diz $ cat conftest.C 
int main () {
        if (sizeof(long long) != 8) { return 1; } 
        return 0;
}
diz $ c++ -pedantic conftest.C
conftest.C: In function `int main()':
conftest.C:2: error: ISO C++ does not support `long long'

This failure to compile causes configure to emit:

checking whether C++ compiler has -pedantic long long bug... yes
configure: error: Your compiler appears to have a known bug where long long is m
iscompiled when using -pedantic.  Reconfigure using --disable-pedantic. 

The reason things have changed is that the GNU compiler, in version 3.3.1, has
changed pedantic warning into errors. In this case the error is that "long long"
is not a standard type, so c++ refuses to compile it.

It seems to me that mozilla's configure test needs to be changed; I'm not sure
how it should be changed, since it's not clear to me exactly what it's testing.
Did the previous GNU bug we're checking for cause (sizeof(long long) != 8) only
when compiled pedantic? Such code will no longer compile, so a configure test
for it wouldn't be required, for c++ >= 3.3.1.

[Note that the workaround is as stated: Reconfigure using --disable-pedantic]
Status: RESOLVED → REOPENED
Resolution: INVALID → ---
Attached patch patchSplinter Review
Add -Wno-long-long before adding -pedantic.
Attachment #129559 - Flags: review?(bryner)
Attachment #129559 - Flags: approval1.5b?
Attachment #129559 - Flags: approval1.4.x?
(The problem is that the test for the long long bug doesn't compile with
-pedantic without -Wno-long-long.)
Attachment #129559 - Flags: review?(bryner) → review+
Comment on attachment 129559 [details] [diff] [review]
patch

a=asa (on behalf of drivers) for checkin to Mozilla 1.5beta and the 1.4 branch.
Attachment #129559 - Flags: approval1.5b?
Attachment #129559 - Flags: approval1.5b+
Attachment #129559 - Flags: approval1.4.x?
Attachment #129559 - Flags: approval1.4.x+
Fix checked in to trunk, 2003-08-11 23:01 -0700.
Fix checked in to MOZILLA_1_4_BRANCH, 2003-08-11 23:05 -0700.
Status: REOPENED → RESOLVED
Closed: 21 years ago21 years ago
Keywords: fixed1.4.1
Resolution: --- → FIXED
Blocks: 224532
Product: Browser → Seamonkey
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: