201766 - Building mozilla stops when compiling nsObjectFrame

Status: VERIFIED WORKSFORME

(SeaMonkey :: Build Config, defect)

Description

[Frederic Bezies]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4b) Gecko/20030411
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4b) Gecko/20030411

All is in the summary.

Trunk source updated with CVS until bugfix for bug 198364.

My .mozconfig :

ac_add_options --enable-calendar
ac_add_options --enable-crypto
ac_add_options --enable-extensions=default,irc
ac_add_options --disable-tests
ac_add_options --disable-debug
ac_add_options --enable-optimize
ac_add_options --enable-strip
ac_add_options --disable-pedantic
ac_add_options --enable-xft



Reproducible: Always

Steps to Reproduce:
1.See details

Actual Results:  
Build is crashing when compiling nsObjectFrame.ccp/.h

Expected Results:  
Build process goes its end.

Related to patches added for fixing bugs 191246 and 196280 yesterday at 9:55 pm ?

Comment 1

[Frederic Bezies]

Additionnal infos : I am using GCC 3.2 (the one with Linux Mandrake 9.1)

gcc -v gave :

Reading specs from /usr//bin/../lib/gcc-lib/i586-mandrake-linux-gnu/3.2.2/specs
Configured with: ../configure --prefix=/usr --libdir=/usr/lib
--with-slibdir=/lib --mandir=/usr/share/man --infodir=/usr/share/info
--enable-shared --enable-threads=posix --disable-checking --enable-long-long
--enable-__cxa_atexit --enable-languages=c,c++,ada,f77,objc,java
--host=i586-mandrake-linux-gnu --with-system-zlib
Thread model: posix
gcc version 3.2.2 (Mandrake Linux 9.1 3.2.2-3mdk)

I will attach build log for error.

Comment 2

[Frederic Bezies]

Attachment: build log error

Comment 3

[Andrew Schultz]

Comment on attachment 120304 [details]
build log error

no error here...
if you go to do mozilla/layout/html/base/src and do "make nsObjectFrame.o" does
it give you anything useful?

note that tinderboxen are doing fine:
http://tinderbox.mozilla.org/showbuilds.cgi?tree=SeaMonkey

Comment 4

[Mitch]

Compiles fine for me on Linux with cvs 20030412 (todays cvs pull)
However i'm using gcc-2.95 (the stable compiler)

Comment 5

[Frederic Bezies]

I know that tinderboxes are green. The worst part is that I grabbed a blank new
tarball source and apply CVS updates to it.

Is gcc 3.2 guilty here ?

I've done your "make", and I had another error, saying there is a type problem...

I don't understand.

Here is the "typical" error message I can see :
"
{standard input}:503: Error: symbol `NS_ENSURE_TRUE' is already defined
{standard input}:509: Error: symbol `NS_ENSURE_TRUE' is already defined
{standard input}:515: Error: symbol `NS_ENSURE_TRUE' is already defined
{standard input}:527: Error: symbol `NS_ENSURE_TRUE' is already defined
{standard input}:533: Error: symbol `t2p' is already defined
{standard input}:539: Error: symbol `NS_ENSURE_TRUE' is already defined
{standard input}:545: Error: symbol `GetOffsetFromView' is already defined
{standard input}:551: Error: symbol `rv' is already defined
{standard input}:575: Error: symbol `NS_ENSURE_TRUE' is already defined
{standard input}:581: Error: symbol `NS_ENSURE_TRUE' is already defined
{standard input}:587: Error: symbol `NS_ENSURE_TRUE' is already defined
{standard input}:599: Error: symbol `NS_ENSURE_TRUE' is already defined
{standard input}:605: Error: symbol `rv' is already defined
{standard input}:611: Error: symbol `NS_ENSURE_SUCCESS' is already defined
{standard input}:617: Error: symbol `child' is already defined
{standard input}:623: Error: symbol `CallQueryInterface' is already defined
{standard input}:647: Error: symbol `QueryInterface' is already defined
{standard input}:667: Error: symbol `NS_ENSURE_SUCCESS' is already defined
{standard input}:691: Error: symbol `NS_ENSURE_ARG_POINTER' is already defined
{standard input}:697: Error: symbol `NS_ENSURE_SUCCESS' is already defined
{standard input}:703: Error: symbol `result' is already defined
{standard input}:709: Error: symbol `NS_RELEASE' is already defined
{standard input}:715: Error: symbol `NS_ADDREF' is already defined
{standard input}:721: Error: symbol `NS_ENSURE_TRUE' is already defined
{standard input}:727: Error: symbol `NS_ENSURE_TRUE' is already defined
{standard input}:733: Error: symbol `NS_ENSURE_TRUE' is already defined
{standard input}:739: Error: symbol `rv' is already defined
{standard input}:745: Error: symbol `rv' is already defined
{standard input}:751: Error: symbol `NS_ENSURE_SUCCESS' is already defined
{standard input}:757: Error: symbol `NS_ENSURE_TRUE' is already defined
{standard input}:769: Error: symbol `NS_ASSERTION' is already defined
{standard input}:775: Error: symbol `rv' is already defined
{standard input}:781: Error: symbol `rv' is already defined
{standard input}:787: Error: symbol `rv' is already defined
{standard input}:793: Error: symbol `rv' is already defined
{standard input}:799: Error: symbol `rv' is already defined
{standard input}:805: Error: symbol `rv' is already defined
{standard input}:811: Error: symbol `rv' is already defined"

A lot of this error.

Any idea ?

Yesterday, building was working without a problem !

Comment 6

[Frederic Bezies]

About comment #4 :

Smells like a gcc 3.2 problem.

Comment 7

[Frederic Bezies]

modifying summary with infos on comment #4.

Comment 8

[Andrew Schultz]

the error messages in comment 5 are from compiling
mozilla/layout/base/src/nsObjectFrame.cpp?  The first mention of t2p is on line
1215.

Is that the *start* of the error messages?  Sometimes the first error will
confuse the compiler and it will be unable to do anything useful with the rest
of the file.

Comment 9

[Andrew Schultz]

compiling nsObjectFrame.cpp (with -pedantic) with gcc 3.2 (cvs updated this
morning):

In file included from nsObjectFrame.cpp:135:
nsObjectFrame.h:184: warning: friend declaration requires class-key, i.e. 
   `friend struct nsPluginInstanceOwner'
nsObjectFrame.cpp:4074: warning: `void ConvertTwipsToPixels(nsIPresContext&, 
   nsRect&, nsRect&)' defined but not used

Comment 10

[Frederic Bezies]

In answer to comment #8, yes it was compiling this file. I cannot see the first
line, because it is too far for my poor xterm.

Comment 11

[Andrew Schultz]

> I cannot see the first line, because it is too far for my poor xterm.

make nsObjectFrame.o |& less -X

Comment 12

[Frederic Bezies]

I am having the same error.

In file included from nsObjectFrame.cpp:135:
nsObjectFrame.h:184: friend declaration requires class-key, i.e. `friend struct
nsPluginInstanceOwner'
nsObjectFrame.cpp:4074: warning: `void ConvertTwipsToPixels(nsIPresContext&,
nsRect&, nsRect&)' defined but not used
make: *** [nsObjectFrame.o] Erreur 1

I also noticed this note from bzbarsky%mit.edu

"Fixing OS/2 bustage -- compiler is too dumb to tell that the void* version is
_not_ what we want here... ;)"

Any relation with this bug ? Or am I completely wrong here ?

Comment 13

[Boris Zbarsky [:bzbarsky]]

> Any relation with this bug ?

Not likely.  If that were the error, gcc would tell you so.

So let me get this straight. gcc errors out without saying what the error is?

Comment 14

[Frederic Bezies]

There is this warning : "warning: `void ConvertTwipsToPixels(nsIPresContext&,
nsRect&, nsRect&)' defined but not used" and after gcc kills mozilla building
process.

Comment 15

[Boris Zbarsky [:bzbarsky]]

That's a warning, not an error.

Does the gcc process crash?  Or dose it just sorta bail out?

Comment 16

[Frederic Bezies]

Yes, build process is completely ended, killed. I posted crash log as an attachment.

After trying to compile nsObjectFrame, gcc stops building mozilla.

Comment 17

[Boris Zbarsky [:bzbarsky]]

That log says absolutely nothing useful -- it looks like a successful compile,
in fact (except for not compiling other things, of course).  My question is,
does gcc crash or just exit with an error condition?

Also, please do the following:

1)  go into mozilla/layout and 'make'
2)  go into mozilla/layout/html/base/src and 'make'

I'd like to see the last bits of output from those two, if they differ from the
already attached log.

Comment 18

[Frederic Bezies]

Attachment: first log wanted

Comment 19

[Frederic Bezies]

Attachment: 2nd log wanted

Comment 20

[Boris Zbarsky [:bzbarsky]]

Those logs look fine.... (apart from the fact that you never built in content/
of course...)

Do you use a parallel build (with -j) or something?

Are things fine if you build content/ then layout/ ?

Comment 21

[Andrew Schultz]

Comment on attachment 120318 [details]
first log wanted

>make[2]: *** Pas de règle pour fabriquer la cible « ../../dist/lib/libgkconbase_s.a », nécessaire pour « libgklayout.so ». Arrêt.

translation:
*** No rule to make target ` ../../dist/lib/libgkconbase_s.a', needed by
`libgklayout.so'.  Stop.

it's acting like mozilla/content was not built (at least completely). 
libgkconbase_s.a should be in mozilla/content/base/src

Comment 22

[Andrew Schultz]

Comment on attachment 120318 [details]
first log wanted

>make[4]: Entering directory `/mozilla/layout/html/base/src'
>nsObjectFrame.cpp
>c++ -o nsObjectFrame.o -c -DOSTYPE=\"Linux2.4.21-0\" -DOSARCH=\"Linux\" -D_IMPL_NS_LAYOUT -I./../../../xul/base/src 
[snip]
>In file included from nsObjectFrame.cpp:135:
>nsObjectFrame.h:184: warning: friend declaration requires class-key, i.e. `friend struct nsPluginInstanceOwner'
>nsObjectFrame.cpp:4074: warning: `void ConvertTwipsToPixels(nsIPresContext&, nsRect&, nsRect&)' defined but not used
>nsPageFrame.cpp
>c++ -o nsPageFrame.o -c -DOSTYPE=\"Linux2.4.21-0\" -DOSARCH=\"Linux\" -D_IMPL_NS_LAYOUT -I./../../../xul/base/src 
[snip]

it looks like nsObjectFrame is compiling successfully.

Comment 23

[Frederic Bezies]

Well, it looks like this warning was the build process killer.

Weird, isn't it ?

You can close this bug as fixed, because my computer had just ended a successful
build process.

Comment 24

[Boris Zbarsky [:bzbarsky]]

Nothing got fixed.... marking worksforme.

Comment 25

[Tracy Walker [:tracy]]

verified