GCC 4.1[beta]

[Gentree]

Sorry, maybe I am missing the point here.

I already had k3b compiled and installed with 3.4.x , only one other K-related pkg displayed this error. emerge sync should pull in the digest as well as the ebuilds.

Usually the need to run digest indicates a corruption.

I dont understand how the choice of compiler version should affect the existance or validity of the manifest.

If I had played with the files or created an overlay I would expect the run ebuild ...digest but in this context it seems inappropriate.

Could you clarify your point?

TIA/.,j

[Tiger683]

hmm, you are right...
but then again, i don't see another option as to
wait until a sync of portage pulls the Manifest file...do you eventually have the feeling your box has been compromised or an FS corruption occured?

[Gentree]

hmm, you are right...
but then again, i don't see another option as to
wait until a sync of portage pulls the Manifest file...do you eventually have the feeling your box has been compromised or an FS corruption occured?

No, not at this stage , I shall look into what happened.



[EDIT] had a closer look and it was an overlay problem .

[Tiger683]

Halcy0n:

gcc-4_1-branch created 30 hours ago, looks like
they're preparing for rc1 or something.....
This might explain lack of a snapshot.....

[Birtz]

Halcy0n:

gcc-4_1-branch created 30 hours ago, looks like
they're preparing for rc1 or something.....
This might explain lack of a snapshot.....

According to homepage http://gcc.gnu.org/ 4.1.0 is released. Still last status is dated 2005-10-10, so it could be "paper release".

[Tiger683]

Current release series: GCC 4.1.0

It only means, that gcc-4.1.0 is currently main development branch....
There have been no releases from it.....

News/Announcements
October 26, 2005
GCC has moved from CVS to SVN for revision control.
September 28, 2005
GCC 4.0.2 has been released.
(..)

[Birtz]

Hm, I disagree, according to homepage:

Current release series: GCC 4.1.0

and

Active development (mainline): will become GCC 4.2.0 (current changes)

But let us not pick on minor word games

[UncleOwen]

Code: Select all

 
ebuild /usr/portage/overlay/app-cdr/k3b/k3b-0.12.1.ebuild digest

or short emerge --digest k3b.

We all know this is a bad thing, unless you have verified that no one tampered with the files.

[Tiger683]

Code: Select all

 
ebuild /usr/portage/overlay/app-cdr/k3b/k3b-0.12.1.ebuild digest

or short emerge --digest k3b.

We all know this is a bad thing, unless you have verified that no one tampered with the files.

has already been cleared in following posts, and someone HAS tampered with them, namely the user himself, because in the path you can see he used an overlay ebuild, which make it needed anyways to generate a new digest.... So let's not be picky, ok?


Birtz: show me gcc-4.1.0 in the releases directory of gcc.gnu.org/pub and i will not argue anymore...

one thing is right thou, gcc-4.1 branch WILL be the next to produce a release, that'S why it's the current release series, and not current development aka. HEAD, aka trunk or whatever you want to call it.....

[Birtz]

Birtz: show me gcc-4.1.0 in the releases directory of gcc.gnu.org/pub and i will not argue anymore...

one thing is right thou, gcc-4.1 branch WILL be the next to produce a release, that'S why it's the current release series, and not current development aka. HEAD, aka trunk or whatever you want to call it.....

Yes, but as I already said, it could be paper release. I only observed the change on the gcc homepage (4.1.0 replacing text which stated 4.0.2 is current release), and that is the only reason why I was lead to beleive that gcc-4.1.0 is actually released, you can't deny that this change is actually there on the homepage. According to CVS though, this change is only cosmetic. And ofcourse, as I have also expressed myself, I don't want to argue on such minor word games

Best regards

[UncleOwen]

Code: Select all

 
ebuild /usr/portage/overlay/app-cdr/k3b/k3b-0.12.1.ebuild digest

or short emerge --digest k3b.

We all know this is a bad thing, unless you have verified that no one tampered with the files.

has already been cleared in following posts,

Sorry, didn't see there was another page...

and someone HAS tampered with them, namely the user himself, because in the path you can see he used an overlay ebuild, which make it needed anyways to generate a new digest.... So let's not be picky, ok?

But you didn't know this, did you?

The thing is: The moment Gentoo (or some mirror) is compromised in some way (let's all hope this does never happen, but we cannot be sure, can we?), one of two things is gonna happen:

• $user goes to forums.gentoo.org, uses search function, sees "Oh, I have to run emerge --digest", and installs compromised software
• $user goes to forums.gentoo.org, does not use search function but creates a new post instead. Some other user posts "run emerge --digest". Same result as above.

That's why it is a _VERY_ bad idea, to advise someone to run emerge --digest without additional information.

[Tiger683]

But you didn't know this, did you?

Yes i did... :

!!! No package manifest found: /usr/portage/overlay/app-cdr/k3b/Manifest

Other than that, i agree with you....post edited

Ok, now let's concentrate on the subject

PS: Birtz: now i get what you meant

[UncleOwen]

But you didn't know this, did you?

Yes i did... :

!!! No package manifest found: /usr/portage/overlay/app-cdr/k3b/Manifest

oops...

Other than that, i agree with you....post edited

Thanks.

Ok, now let's concentrate on the subject

Great idea

[Tiger683]

Looks like next snapshot will come later.
The cronjob ahs been disabled because of branch creation,
which means it can take a while. Final changes were done just about an hour ago, so
it could be tomorrow.... argh.... im making my own snapshot i guess.....

[Gentree]

Glad to see all that's cleared up.

good job that a new release is just around the corners. I have a rather long list of pkgs that refuse to compile with 4.1-beta:

Code: Select all

resumé of "unfinished jobs" breaks:

opensp openjade avifile windowmaker libjsw-1.5.5 tunepimp amarok swig xfe eix 
 smegpeg simgear rosegarden flux fam p7zip mysql

Other breaks:
transcode libquicktime windowmaker ardour qt and kde

I've tried installing 4.0.2 but no joy , the errors are different but the result much the same.

Retro to 3.4.3 seems even less productive.

What versions of qt and kde are meant to be compilable with beta?

Thx.

[EDIT] compiling qt-4.0.1 with gcc 4.1.0 , looking good.
Tried to follow your post higher up by unmasking kde-3.5_rc1 but seems to want all the split stuff unmasking as well. Is that what you built?

TIA

Also apache2 rebuilt ok but wont run:
* Apache2 has detected a syntax error in your configuration files:
Syntax error on line 6 of /etc/apache2/modules.d/70_mod_php.conf:
Cannot load /usr/lib/apache2/modules/libphp4.so into server: libmysqlclient.so.12: cannot open shared object file: No such file or directory

If I try rebuilding php or mod_php I get the following:

Code: Select all

checking for sys/utsname.h... yes
checking for sys/ipc.h... yes
checking for dlfcn.h... yes
checking for fopencookie... yes
configure: error: can not run test program while cross compiling

Well it's hard not to get cross compiling with all these bloody errors

[Tiger683]

gentree: it worked just fine with dev-lang/php, which deprecates php-4* and mod_php anyways....
PS: i have extracted quite a few patches from gcc mailing list,
and because there seems to be some breakage in svn, i will prepare an overlay with these patches,
they fix 5 PR targets, bootstrap on x86_64 and (hopefully) the wxWindows issue, which seems to
be a silent ICE, or a memory alloc bug, which it states sometimes here.
I tried to hunt down a testcase for the ML, but no luck for now......

[Tiger683]

Ok, could anyone test this one parallelly?
It should fix a few issues, i'm currently compiling it too....
It contains 10 custom patches, those most interesting for
regression fixing, it appears that the x86_64 bootstrap breaking patch
wasn't committed yet in 20051112..

gcc-4.1.0_beta20051112-r1_overlay.tar.bz2

[seren]

trying this right now, post your emerge --info, i was wondering what version of binutils glibc you are using.

[luddite]

I've got a problem trying to update to the GCC 4.1 beta ebuild. I'm hoping that somebody can help me.

I've successfully gotten past the typical problems with adding the declaration regarding bug reports to make.conf and adding -ffriend-injection to the CXXFLAGS. I'm running into a problem emerging GCC 4.1:

The package appears to emerge fine, but issues the following error after the emerge:

Code: Select all

--- !targe sym /usr/bin/i686-pc-linux-gnu-gcc-4.1.0-beta20051112
--- !targe sym /usr/bin/i686-pc-linux-gnu-g++-4.1.0-beta20051112
--- !targe sym /usr/bin/i686-pc-linux-gnu-c++-4.1.0-beta20051112
--- !targe sym /usr/bin/gfortran-4.1.0-beta20051112
--- !targe sym /usr/bin/gcc-4.1.0-beta20051112
--- !targe sym /usr/bin/g++-4.1.0-beta20051112
--- !targe sym /usr/bin/c++-4.1.0-beta20051112
>>> original instance of package unmerged safely.
/usr/bin/gcc-config: line 632: /etc/env.d/gcc/i686-pc-linux-gnu-3.4.4: No such file or directory
 * /usr/bin/gcc-config: Profile does not exist or invalid setting for /etc/env.d/gcc/i686-pc-linux-gnu-3.4.4
 * Switching native-compiler to i686-pc-linux-gnu-3.4.4 ...
/usr/bin/gcc-config: line 182: /etc/env.d/gcc/i686-pc-linux-gnu-3.4.4: No such file or directory
gawk: cmd. line:1: fatal: cannot open file `/etc/env.d/gcc/i686-pc-linux-gnu-3.4.4' for reading (No such file or directory  [ ok ]

 * If you have issues with packages unable to locate libstdc++.la,
 * then try running 'fix_libtool_files.sh' on the old gcc versions.

 * This gcc-4 ebuild is provided for your convenience, and the use
 * of this compiler is not supported by the Gentoo Developers.
 * Please file bugs related to gcc-4 with upstream developers.
 * Compiler bugs should be filed at http://gcc.gnu.org/bugzilla/
>>> Regenerating /etc/ld.so.cache...
>>> sys-devel/gcc-4.1.0_beta20051112 merged.

>>> clean: No packages selected for removal.

>>> Auto-cleaning packages ...

>>> No outdated packages were found on your system.


 * Regenerating GNU info directory index...
 * Processed 6 info files.

The problem appears to be that GCC is successfully emerged, but gcc-config appears to be making mistakes at lines 632 and 182 where it assumes that I am upgrading from GCC 3.4.4. I am not. I am upgrading from GCC 4.0.2 and GCC 3.4.4 was trimmed from my system long ago, when I had first emerged GCC 4.0. My GCC 4.0 toolkit has been working fine, but I can't seem to emerge GCC 4.1.

When I look at the output of gcc-config, it is obvious that no GCC compiler has been chosen. This is interestting, as I had previously had GCC 4.0 properly configured via gcc-config, and it seems to have been de-configured during the process of emerging GCC 4.1 beta:

Code: Select all

# gcc-config -l
 [1] i686-pc-linux-gnu-4.0.2
 [2] i686-pc-linux-gnu-4.1.0-beta20051112

I can successfully set the compiler using "gcc-config 2", but as soon as I perform an "emerge -e system" the gcc-config setting is lost as soon as GCC gets re-emerged. When it gets re-emerged, GCC issues the same errors that Ihad mentioned previously, and gcc-config shows that no compiler is chosen, etc. So I choose "gcc-config 2" again and try to re-emerge and the process repeats itself. I get stuck in an endless loop.

Can anyone tell me how I can fix this problem?

[Tiger683]

seren: binutils 2.16.91.0.4, glibc 2.3.90 (snapshot 20051120 with 20051115 fedora tarball), gcc from above overlay, so 4.1.0-20051112 with the patches...

luddite: tried emerge gcc-config ?
I had some issues with gcc-config too after upgrading from 4.0.2 at my very first try with 20051105,
but for me it was a little different error.
i did gcc-config 2. emerge --oneshot gcc-config, gcc-config -l, which returned right compiler,
fix_libtool_files.sh 4.0.2 --oldarch i686-pc-linux-gnu

then i pruned the old compiler.....

[thersites]

luddite - i seem to remember that cleaning out some of the old files from /etc/env.d/gcc cleared up that gcc-config problem for me
-- but make backups!!

[Gentree]

Ok, could anyone test this one parallelly?
It should fix a few issues, i'm currently compiling it too....
It contains 10 custom patches, those most interesting for
regression fixing, it appears that the x86_64 bootstrap breaking patch
wasn't committed yet in 20051112..

gcc-4.1.0_beta20051112-r1_overlay.tar.bz2

compiling now, any need to reuild the toolchain/system/wholedamnworld again?

It's already been through the mixer with beta20051112

I guess emerge -e system would be a good idea.

Thanks for the overlay, nice and easy.

[Tiger683]

Ok, could anyone test this one parallelly?
It should fix a few issues, i'm currently compiling it too....
It contains 10 custom patches, those most interesting for
regression fixing, it appears that the x86_64 bootstrap breaking patch
wasn't committed yet in 20051112..

gcc-4.1.0_beta20051112-r1_overlay.tar.bz2

compiling now, any need to reuild the toolchain/system/wholedamnworld again?

It's already been through the mixer with beta20051112

I guess emerge -e system would be a good idea.

Thanks for the overlay, nice and easy.

emerge -e system should be enough, can you then tell me if wxGTK still breaks with this one?

[seren]

any chance you can put that version of glibc 2.3.9 in the overlay?

[Gentree]

system is already underway.

I did manage to get a few more packages compiled. gtk+ did compile first time around but several packages were complaining about gtk or gtk-config.

It seems that once I gave up on --as-needed a few of these cheered up and started feeling a whole lot better.

I already have bitwatcher , mplayer , more importantly wxGTK rebuilt on the portage 4.1beta.

Swapping to 4.0.2 did not seem to do much for anything. I tried serveral of the sticky ones and did not get any further.

I have great hopes for you new r1.

I will post what I get.

BTW one thing that makes me very sad is that conky has conked out. Just segfaults right off the bat. Not a life-shattering event but it is a useful tool.