Gnome-2.6, 5 pkgs fail to compile!

[wrc1944]

Just emerged gnome 2.6.0, and got 44 of 52 packages compiled using my standard cflag set (-O3 -aggresive). 8 failed, but after dropping cflags to:

-O2 -march=athlon-xp -pipe -fomit-frame-pointer, 3 more installed (eel, nautilis, and nautilis cd-burner).

These five still fail to compile:

Exiting: U!!! ERROR: gnome-base/libgnomeprint-2.6.0 failed.
!!! Function gnome2_src_compile, Line 39, Exitcode 2
!!! compile failure

!!! ERROR: gnome-base/libgnomeprintui-2.6.0 failed.
!!! Function econf, Line 365, Exitcode 1
!!! econf failed

!!! ERROR: media-gfx/eog-2.6.0 failed.
!!! Function econf, Line 365, Exitcode 1
!!! econf failed

!!! ERROR: app-text/gpdf-0.131 failed.
!!! Function gnome2_src_compile, Line 39, Exitcode 2
!!! compile failure

!!! ERROR: app-editors/gedit-2.6.0 failed.
!!! Function econf, Line 365, Exitcode 1
!!! econf failed

The last two seem to depend on libgnomeprint. Anybody experiencing this, or have a solution? I really need gedit and gpdf. I can post more lines of the errors, if need be. I searche around, and found some references to needing dev-perl/XML-Parser to solve !!! Function econf, Line 365, Exitcode 1, but I checked, and I already have the latest version.

Hmmmm- just thought- does the printer have to be on when compiling the printing packages? I wouldn't think so, but maybe it does?

wrc1944
_________________
Main box- AsRock x370 Gaming K4
Ryzen 7 3700x, 3.6GHz, 16GB GSkill Flare DDR4 3200mhz
Samsung SATA 1000GB, Radeon HD R7 350 2GB DDR5
OpenRC Gentoo ~amd64 plasma, glibc-2.36-r7, gcc-13.2.1_p20230304
kernel-6.9.1 USE=experimental python3_11

[dizzogg]

i'm sure that the printer doesn't need to be on when compiling printing packages, but I'm not too sure why those are failing.... Can you give us more info as to the error messages you're getting?

[Vann]

The actual errors are in the few lines above those, so those don't really give us much information. Could you post those lines?

[wrc1944]

Here's the output- you can see what my standard flags are, but all these also failed after dropping to the ones mentioned above. This is curious, because gnome-2.4 compiled perfectly with even more agresive flags than my current standard set. When I try and compile gedit or gpdf separately, it begins with libgnomeprint (1 of 3), and of course errors out.

My PKG_CONFIG_PATH environment should be standard, as I've never messed with it since I installed Gentoo. Anyway, since the packages requested aren't installed, and thus couldn't be found whatever thePKG_CONFIG_PATH environment was , how could that be the problem?

wrc1944
------------------------------------------------------------------
gcc -DHAVE_CONFIG_H -I. -I. -I../.. -I../.. -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -I/usr/include/libart-2.0 -I/usr/include/pango-1.0 -I/usr/include/libxml2 -I/usr/include/freetype2 -DGNOMELOCALEDIR=\"/usr/share/locale\" -march=athlon-xp -O3 -pipe -fomit-frame-pointer -ftracer -ffast-math -fprefetch-loop-arrays -Wall -Wmissing-prototypes -Wnested-externs -Wpointer-arith -Wno-sign-compare -c gp-transport-lpr.c -MT gp-transport-lpr.lo -MD -MP -MF .deps/gp-transport-lpr.TPlo -fPIC -DPIC -o .libs/gp-transport-lpr.o
{standard input}: Assembler messages:
{standard input}:79: Error: invalid character '_' in mnemonic
{standard input}:121: Error: invalid character '@' in mnemonic
{standard input}:140: Error: invalid character '_' in mnemonic
{standard input}:178: Error: invalid character '_' in mnemonic
{standard input}:198: Error: invalid character '@' in mnemonic
{standard input}:209: Error: invalid character '@' in mnemonic
{standard input}:220: Error: invalid character '@' in mnemonic
{standard input}:234: Error: invalid character '_' in mnemonic
{standard input}:253: Error: invalid character '@' in mnemonic
{standard input}:279: Error: invalid character '_' in mnemonic
{standard input}:294: Error: invalid character '_' in mnemonic
{standard input}:295: Error: suffix or operands invalid for `lea'
{standard input}:295: Error: invalid character '_' in mnemonic
{standard input}:297: Error: suffix or operands invalid for `lea'
{standard input}:297: Error: invalid character '_' in mnemonic
{standard input}:298: Error: invalid character '@' in mnemonic
{standard input}:299: Error: suffix or operands invalid for `lea'
{standard input}:299: Error: invalid character '_' in mnemonic
{standard input}:301: Error: suffix or operands invalid for `lea'
{standard input}:301: Error: invalid character '_' in mnemonic
{standard input}:304: Error: suffix or operands invalid for `lea'
{standard input}:304: Error: invalid character '_' in mnemonic
{standard input}:334: Error: invalid character '_' in mnemonic
{standard input}:347: Error: suffix or operands invalid for `mov'
{standard input}:347: Error: invalid character '@' in mnemonic
{standard input}:350: Error: invalid character '_' in mnemonic
{standard input}:370: Error: invalid character '@' in mnemonic
{standard input}:373: Error: invalid character '@' in mnemonic
{standard input}:392: Error: invalid character '_' in mnemonic
make[3]: *** [gp-transport-lpr.lo] Error 1
make[3]: *** Waiting for unfinished jobs....
gcc -DHAVE_CONFIG_H -I. -I. -I../.. -I../.. -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -I/usr/include/libart-2.0 -I/usr/include/pango-1.0 -I/usr/include/libxml2 -I/usr/include/freetype2 -DGNOMELOCALEDIR=\"/usr/share/locale\" -march=athlon-xp -O3 -pipe -fomit-frame-pointer -ftracer -ffast-math -fprefetch-loop-arrays -Wall -Wmissing-prototypes -Wnested-externs -Wpointer-arith -Wno-sign-compare -c gp-transport-file.c -MT gp-transport-file.lo -MD -MP -MF .deps/gp-transport-file.TPlo -fPIC -DPIC -o .libs/gp-transport-file.o
gcc -DHAVE_CONFIG_H -I. -I. -I../.. -I../.. -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -I/usr/include/libart-2.0 -I/usr/include/pango-1.0 -I/usr/include/libxml2 -I/usr/include/freetype2 -DGNOMELOCALEDIR=\"/usr/share/locale\" -march=athlon-xp -O3 -pipe -fomit-frame-pointer -ftracer -ffast-math -fprefetch-loop-arrays -Wall -Wmissing-prototypes -Wnested-externs -Wpointer-arith -Wno-sign-compare -c gp-transport-file.c -MT gp-transport-file.lo -MD -MP -MF .deps/gp-transport-file.TPlo -o gp-transport-file.o >/dev/null 2>&1
make[3]: Leaving directory `/var/tmp/portage/libgnomeprint-2.6.0/work/libgnomeprint-2.6.0/libgnomeprint/transports'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory `/var/tmp/portage/libgnomeprint-2.6.0/work/libgnomeprint-2.6.0/libgnomeprint'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/var/tmp/portage/libgnomeprint-2.6.0/work/libgnomeprint-2.6.0'
make: *** [all] Error 2

!!! ERROR: gnome-base/libgnomeprint-2.6.0 failed.
!!! Function gnome2_src_compile, Line 39, Exitcode 2
!!! compile failure

(gnome-base/eel-2.6.0 had a similar error, but dropping cflags fixed it).
----------------------------------------------------------------------

checking for catalogs to be installed... am ar az be bg bn ca cs cy da de el en_CA en_GB es et eu fa fi fr ga gl he hi hr hu id is it ja kn ko li lt lv ml mn ms nl nn no pa pl pt pt_BR ro ru sk sl sq sr sr@Latn sv ta th tr uk vi wa zh_CN zh_TW
checking for pkg-config... /usr/bin/pkg-config
checking for
gtk+-2.0 >= 2.4.0
libgnomeprint-2.2 >= 2.6.0
libgnomecanvas-2.0 >= 1.117.0
gnome-icon-theme >= 1.1.92
... Requested 'libgnomeprint-2.2 >= 2.6.0' but version of libgnomeprint-2.2 is 2.4.2

configure: error: Library requirements (
gtk+-2.0 >= 2.4.0
libgnomeprint-2.2 >= 2.6.0
libgnomecanvas-2.0 >= 1.117.0
gnome-icon-theme >= 1.1.92
) not met; consider adjusting the PKG_CONFIG_PATH environment variable if your libraries are in a nonstandard prefix so pkg-config can find them.

!!! ERROR: gnome-base/libgnomeprintui-2.6.0 failed.
!!! Function econf, Line 365, Exitcode 1
!!! econf failed
---------------------------------------------------------------------------------------

checking for jpeg_simple_progression in -ljpeg... yes
checking for libgnomeprintui-2.2 >= 2.5.5 gconf-2.0 >= 2.5.90 libglade-2.0 >= 2.3.6 libgnomecanvas-2.0 >= 2.5.92... Requested 'libgnomeprintui-2.2 >= 2.5.5' but version of libgnomeprintui-2.2 is 2.4.2

configure: error: Library requirements (libgnomeprintui-2.2 >= 2.5.5 gconf-2.0 >= 2.5.90 libglade-2.0 >= 2.3.6 libgnomecanvas-2.0 >= 2.5.92) not met; consider adjusting the PKG_CONFIG_PATH environment variable if your libraries are in a nonstandard prefix so pkg-config can find them.

!!! ERROR: media-gfx/eog-2.6.0 failed.
!!! Function econf, Line 365, Exitcode 1
!!! econf failed
-------------------------------------------------------------------------

gpdf-control.o -MD -MP -MF ".deps/gpdf-control.Tpo" \
-c -o gpdf-control.o `test -f 'gpdf-control.cc' || echo './'`gpdf-control.cc; \
then mv -f ".deps/gpdf-control.Tpo" ".deps/gpdf-control.Po"; \
else rm -f ".deps/gpdf-control.Tpo"; exit 1; \
fi
gpdf-control.cc: In function `void gpdf_print_job_prepare(GPdfPrintJob*)':
gpdf-control.cc:712: error: `GNOME_PRINT_KEY_DUPLEX' undeclared (first use this
function)
gpdf-control.cc:712: error: (Each undeclared identifier is reported only once
for each function it appears in.)
make[4]: *** [gpdf-control.o] Error 1
make[4]: *** Waiting for unfinished jobs....
make[4]: Leaving directory `/var/tmp/portage/gpdf-0.131/work/gpdf-0.131/xpdf'
make[3]: *** [all-recursive] Error 1
make[3]: Leaving directory `/var/tmp/portage/gpdf-0.131/work/gpdf-0.131/xpdf'
make[2]: *** [all] Error 2
make[2]: Leaving directory `/var/tmp/portage/gpdf-0.131/work/gpdf-0.131/xpdf'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/var/tmp/portage/gpdf-0.131/work/gpdf-0.131'
make: *** [all] Error 2

!! ERROR: app-text/gpdf-0.131 failed.
!!! Function gnome2_src_compile, Line 39, Exitcode 2
!!! compile failure
------------------------------------------------------------------------

checking for new_aspell_speller in -laspell... yes
checking for pkg-config... /usr/bin/pkg-config
checking for libgnomeui-2.0 >= 2.5.0 libglade-2.0 >= 2.0.0
eel-2.0 >= 2.5.0 gtksourceview-1.0 >= 0.9.0
libgnomeprintui-2.2 >= 2.5.0... Requested 'eel-2.0 >= 2.5.0' but version of eel is 2.4.2

configure: error: Library requirements (libgnomeui-2.0 >= 2.5.0 libglade-2.0 >= 2.0.0
eel-2.0 >= 2.5.0 gtksourceview-1.0 >= 0.9.0
libgnomeprintui-2.2 >= 2.5.0) not met; consider adjusting the PKG_CONFIG_PATH environment variable if your libraries are in a nonstandard prefix so pkg-config can find them.

!!! ERROR: app-editors/gedit-2.6.0 failed.
!!! Function econf, Line 365, Exitcode 1
!!! econf failed
-----------------------------------------------------------------
_________________
Main box- AsRock x370 Gaming K4
Ryzen 7 3700x, 3.6GHz, 16GB GSkill Flare DDR4 3200mhz
Samsung SATA 1000GB, Radeon HD R7 350 2GB DDR5
OpenRC Gentoo ~amd64 plasma, glibc-2.36-r7, gcc-13.2.1_p20230304
kernel-6.9.1 USE=experimental python3_11

[wrc1944]

Another problem solved!

Apparently, after some more experiments with cflags, I discovered -fomit-frame-pointer was preventing libgnomeprint, libgnomeprintui, media-gfx/eog-2.6.0, app-text/gpdf-0.131, and app-editors/gedit-2.6.0 from compiling. Removing it allowed all 5 to complete without a hitch.

A good lesson learned the hard way (aren't they all!):

You can go along on ~xf86 systems for months with aggressive cflags, and 99% of the time you will have no problems. But when a compile fails, and the error output seems complicated, the problem can be as simple as one cflag conflicting with one or more packages.

I'm going to keep my standard set of aggressive flags in make.conf, but when a compile fails, the first thing I'm trying is a recompile with:

-O2 -march=athlon-xp -pipe.

If it then works, I'll know it's not some other problem, and not waste hours trying to figure out error output messages. Of course if the error message is really obvious, even to an amateur like me, I'll try that first, but it takes a lot less time to change flags, and recompile a package than to search the forum and internet for clues about specific errors, and will probably resolve the issue one way, or the other. If -O2 -march=athlon-xp -pipe fails, then you can always explore other avenues.

wrc1944
_________________
Main box- AsRock x370 Gaming K4
Ryzen 7 3700x, 3.6GHz, 16GB GSkill Flare DDR4 3200mhz
Samsung SATA 1000GB, Radeon HD R7 350 2GB DDR5
OpenRC Gentoo ~amd64 plasma, glibc-2.36-r7, gcc-13.2.1_p20230304
kernel-6.9.1 USE=experimental python3_11

[seltrus]

I compiled gnome 2.6 last night without any trouble with:

[wrc1944]

seltrus,
-Os was my next move, if removing -fomit-frame-pointer failed.

Two questions come to mind. One, since -fomit-frame-pointer worked with a Pentium4, is there an inherit difference in AMD and Intel cpus that would cause this, and two, does -Os with -fomit-frame-pointer produce faster code than -O2 without -fomit-frame-pointer? I know it's nit-picking, whether one or the other produces a 1% performance benefit, but that's part of the fun of Gentoo. Guess I'll just be happy I finally got a full compile of Gnome 2.6.0. I'm really a kde fan, but I'll give the new Gnome a fair trial.

About the errors: Can anyone enlighten me as to what is the general basic cause of such errors like this?

{standard input}:294: Error: invalid character '_' in mnemonic
{standard input}:295: Error: suffix or operands invalid for `lea'

I've been running into these types quite a bit in the last month, in multiples of 100-200 on a borked compile. Obviously, dropping the cflags solves some of them, but I'd like to understand why this is happening, i.e., the root cause, if there is one. I'm thinking it might be a gcc/gtk problem, considering I'm running the latest versions. ( well, not gcc 3.4 as yet- maybe I'll take the plunge on my test Duron 1300 box soon, along with reiser4- I'm waiting until I see a post that somebody has actually done it).

wrc1944
_________________
Main box- AsRock x370 Gaming K4
Ryzen 7 3700x, 3.6GHz, 16GB GSkill Flare DDR4 3200mhz
Samsung SATA 1000GB, Radeon HD R7 350 2GB DDR5
OpenRC Gentoo ~amd64 plasma, glibc-2.36-r7, gcc-13.2.1_p20230304
kernel-6.9.1 USE=experimental python3_11

[MrJ412]

I highly suggest you do some testing of your hardware ... Run memtest86+ overnight ... I was having these problems a last month, and i discovered I was having memory errors ...

After some further investigation and trial and error hardware replacement, I discovered that my CPU was failing ... Luckily it was under warranty and AMD replaced it promptly ... I have been compiling fine since ...

So, I would highly suggest you investigate that, escpecially if you are getting weird errors noone else is getting. Compiles will expose problems not seen during normal use of a pc.

[wrc1944]

Hmmm- I guess checking the memory is worth a shot, however, I run only top quality Crucial DDR, and have never had a Crucial stick fail in 6 years, on many boxes I have built. My cpus & boards are running extremely cool, and have never given my any problems. It's been my experience that cpus either work, or they don't, and the only thing that causes random errors with them is overheating, which is definitely not the case here.

Maybe I mislead somewhat in my previous post- my compile errors are only on specific new updated packages. The vast majority of my compiles are completing fine, with no problems. And, these are on pure ~xf86 systems, so I have to expect problems from time-to-time.

When I said "in multiples of 100-200" I meant the number of lines of these type right before a compile errors out, not the number of individual compiling failures on different packages:

{standard input}:294: Error: invalid character '_' in mnemonic
{standard input}:295: Error: suffix or operands invalid for `lea'

Sorry if my wording was misleading. If I were having failures on 100 random packages, I would certainly suspect hardware, but all the failures I have had have been resolved by cflag/USE adjustments, and/or editing ebuild or other files mentioned in Gentoo Bugzilla, or waiting for a corrected version of the specific package or ebuild.

wrc1944
_________________
Main box- AsRock x370 Gaming K4
Ryzen 7 3700x, 3.6GHz, 16GB GSkill Flare DDR4 3200mhz
Samsung SATA 1000GB, Radeon HD R7 350 2GB DDR5
OpenRC Gentoo ~amd64 plasma, glibc-2.36-r7, gcc-13.2.1_p20230304
kernel-6.9.1 USE=experimental python3_11

[seltrus]

-Os is supposed to be slower than -O2 when the code is actually running. There were some benchmarks on the net exploring this. However, start times are reduced because the binary is smaller which is huge for my slow I/O laptop that needs apps to load quickly to take notes, etc.

I had some trouble with -Os on gaim and gweled even on pentium4. in gweled, i posted a bug report and the developer discovered that it was his code that was broken. In gaim, I believe, -Os is filtered and replaced with -O2.

I doubt it is your hardware, but I have seen some really odd problems crop up with bad memory, so I wouldn't discount it.

My money would be on the fact that Gnome 2.6 sources are not perfect for the first release, and something breaks when gcc optimizes them with -fomit-frame-pointer on -march=athlon-xp.

[MrJ412]

LOL ... ok, i did read that as 100's of packages

I had no problem compiling all packages with "-march=athlon-xp -O3 -pipe -fomit-frame-pointer" ... However, the only packages from ~x86 are the gnome packages.

Oh, and I was under that impression about CPU's as well, which is why I replaced memory and the motherboard before I swapped out the processor and discovered it was the source of the errors. But, it does sound like its probably related to using newer versions of the compiler tools.

[duderonomy]

Glad I found this thread...

I have hit a similar issue with gedit-2.6.0 failing during an
emerge gnome. Previous to that failure I also hit the same
issue with dev-perl/XML-Parser. And like the OP, emerge -s
indicated XML-Parser was already installed.

Due to my n00bness I did not know that I needed to
invoke: emerge --oneshot XML-Parser. The "--oneshot" switch
was kindly pointed out to me (in very short order) by
robmoss2k. Very cool.

I am also on AMD (athlon XP). My make.conf also includes the
-fomit-frame-pointer compiler option. My "emerge gnome" may
have made it further in the number of packages successfully
compiled than the situation expressed by the original poster. For
example, in the case of XML-Parser that merely needed to have
the emerge --oneshot.

Hm... So will emerge --pretend indicate the exact order of the
packages to be built? That would help me determine where I'm
at in the process of emerge gnome.

Has it been determined that it is not compatible to use "-fomit-frame-pointer" on gentoo?

Here are the pertinent options from my make.conf file:

[duderonomy]

Removal of the -fomit-frame-pointer compile option yielded a
similar result. Failed again during the gedit-2.6.0 build.

Can anyone suggest an approach to get through an
"emerge gnome", at this point?

[Ludwig]

I just had to face this problem today...

In fact, there are a lot of threads dealing about this problem but it's hard to extract the solution...

Anyway, you might have upgraded your version of GCC from 3.3.2 to 3.3.3 I guess...

So, looking at the errors you got, look at this script : fix_libtool_files.sh

And try to run this :

fix_libtool_files.sh 3.3.2 --oldarch i386-pc-linux-gnu

It should fix your problem...

Ludwig

[duderonomy]

I am wanting to learn more about this gcc issue...
Also if someone could give me some advice about how to "rebuild" gcc.
Note that my arch is athlon xp, which is 686 instructions...
What is this about "fix_libtool_files.sh 3.3.2 --oldarch i386-pc-linux-gnu "
and 386? Any reference material I should read to educate myself on this?
Can anyone elaborate on the point made by the previous poster?

THX
D