palemoon-33.7.0.linux-x86_64-sse2_gtk2 binary no longer starts with my old Linux Topic is solved

[Lycanthrope_végétarien]

$/opt/palemoon/palemoon
XPCOMGlueLoad error for file /opt/palemoon/libxul.so:
/usr/lib64/libstdc++.so.6: version `CXXABI_1.3.11' not found (required by /opt/palemoon/libxul.so)
Couldn't load XPCOM.

[vannilla]

Your version of glibc is likely too old.
Please post informations about your system, even just the distro name and version if applicable.

[Bababo26]

libstdc++ is a gcc library - the change to using gcc 9.1 is probably causing the problem as it supports CXXABI_1.3.12 (see https://gcc.gnu.org/onlinedocs/libstdc++/manual/abi.html).

[Lycanthrope_végétarien]

My problem came from GCC, thanks Bababo26 for the link https://gcc.gnu.org/onlinedocs/libstdc+ ... l/abi.html
I updated GCC 6.5 ---> 9.5

If I have the same problem with the next Palemoon release I would definitely go to GCC 9.5.

There are a few packages to rebuild:
$/media/DataHita2To/Linux/slack-revdep.sh
OpenBLAS-0.3.29-x86_64-1_SBo
/usr/lib64/libopenblas_nehalemp-r0.3.29.so
libgfortran.so.3 => not found

amd-2.4.6-x86_64-7_SBo
/usr/lib64/libamdf77-2.4.6.so
libgfortran.so.3 => not found

gettext-tools-0.19.8.1-x86_64-1
/usr/lib64/gettext/gnu.gettext.DumpResource
libgcj.so.17 => not found

hdf5-1.14.1-x86_64-1_SBo
/usr/lib64/libhdf5_fortran.so.310.1.0
libgfortran.so.3 => not found

lapack-3.12.1-x86_64-1_SBo
/usr/lib64/liblapack.so.3.12.0
libgfortran.so.3 => not found

octave-6.4.0-x86_64-3_SBo
/usr/bin/octave-cli-6.4.0
libgfortran.so.3 => not found

openmpi-5.0.6-x86_64-1_SBo
/usr/lib64/libmpi_mpifh.so.40.40.1
libgfortran.so.3 => not found

sox-14.4.2-x86_64-9
/usr/bin/sox
libssp.so.0 => not found

[jobbautista9]

I wonder if the system requirements page should be updated. The minimum version of libstdc++ listed there is 4.6.1, which suggests GCC 3.1 according to the ABI policy above...

[Moonchild]

Happy to update it with more accurate values.
If I read this correctly, apparently we should ask for:

• glibc 3.4.26
• libstdc++ 6.0.26

Let me know if this is correct?

[jobbautista9]

libstdc++ version looks correct, but glibc is separate from the C++ toolchain. The latest major version of glibc has not even reached version 3 yet (the latest is 2.41)... Assuming we're still building with CentOS 7 for Linux binaries the minimum glibc should be 2.17.

[Nuck-TH]

Oracle Linux 8 with gcc-toolset-11

[Moonchild]

ok, just going by the documentation on that specific page is wrong. why does Linux have to be so difficult?

[frostknight]

. Assuming we're still building with CentOS 7 for Linux binaries the minimum glibc should be 2.17.

Isn't CentOS 7 deprecated now? It has older binaries I think than even Hyperbola if I recall correctly.

[Moonchild]

Isn't CentOS 7 deprecated now?

It is, and we stopped using it for official builds.

[frostknight]

It is, and we stopped using it for official builds.

Also, even more than that, CentOS 8 is also deprecated and that was made after CentOS 7. CentOS Stream is the current CentOS.

Off-topic:
I read this in wikipedia how CentOS stable was being discontinued, I recall when it happened, it pissed people off

[frostknight]

ok, just going by the documentation on that specific page is wrong. why does Linux have to be so difficult?

Off-topic:
Btw, is BSD any easier than Linux to develop for?

Mainly talking about FreeBSD, NetBSD and OpenBSD and if any others, DragonFlyBSD.

I hear that DragonFlyBSD and OpenBSD are constantly changing. I imagine that annoys people lol.

[Mike_Walsh]

ok, just going by the documentation on that specific page is wrong. why does Linux have to be so difficult?

Heh. Tell me, Moonchild; you don't by any chance belong to the camp that firmly believes it would be so much simpler - for everyone! - if ALL Linux distros were to combine into a single distro.....just like there's a single version of Windows, or a single version of MacOS.....so as to make devs lives easier?

I'm genuinely curious.....'cos there's 'pros' & 'cons' to everything in this world.

=====================================

Lennart Poettering has been pushing for this for a LONG time.....but the individual projects don't want to know. He seriously believes that every distro should give up its individuality, and all those tens of thousands of people should merge their code/projects/individual ideas purely for the good of the entire community. Mind you, Poettering ultimately wants Red Hat to control everything.....which would mean IBM controlling everything. Which would take us back, full-circle (again), to what the landscape was like 50 years ago....when IBM literally 'ruled the roost'. Those guys have LONG memories.....and they've never forgiven Microsoft for the way they f**ked them over with regard to the DOS fiasco.

If they could 'use' the entire Linux community as a way to browbeat Microsoft, they'd be well content.

(*shrug...*)


Mike.

[Moonchild]

Off-topic:

Tell me, Moonchild; you don't by any chance belong to the camp that firmly believes it would be so much simpler - for everyone! - if ALL Linux distros were to combine into a single distro

No, I can't say that I do. But I do firmly believe that there are way too many distros out there, one for every dev who wants an OS to be catered 100% and no less to their own workflow, give or take. Freedom of choice is working against Linux's potential.
Within the Linux community I also see a divide between people who want to have this personal catering for themselves, and people who have a conviction that they want to force their way of working on everyone else. In the latter group also fall the commercial variants of Linux that are IMHO way too top-down oriented, resulting in part in the reactionary former group who would rather have a distro used by 100 people and being on their own than to bend to that commercial edge.

These forces keep a strong divide within the Linux sector, and prevents cooperation and constructive feedback.

[Mike_Walsh]

Off-topic:
Hm. That's interesting.

I felt sure that the original premise I postulated would appeal to someone like yourself as a developer, since it would no doubt simplify your life to quite an extent. There again, the individual distro requirements should be taken care of by the requisite compiler/environment/toolchain, etc. If I'm right, that shouldn't actually alter what you're doing vis-a-vis the way the browser itself functions.

.....or do I have that wrong?

Mike.

(BTW, I agree there's far too many distros out there. This is why I think the Puppy approach is better, since every Puppy ever created has contained a full suite of re-mastering tools. You can take any Puppy 'base' you want (built around whatever your preferred mainstream 'base' happens to be), customize it to suit yourself, set it up for specialised operation, the sky's the limit.....then use the re-mastering tools - which are fully automated, after entering a few simple parameters - to produce an "immutable" copy.....which will always boot up into the individual's preferred environment/setup.

If the individual wishes to "share" their specialist creation, they can post in the community's Forum. We call these "Puplets".)

Myself, I think re-mastering doesn't get anywhere near the amount of coverage it really deserves.

[Moonchild]

Off-topic:

...or do I have that wrong?

You have it wrong.
The issue isn't compiling from source itself, but rather how different distros handle their system libraries/drivers/modules/kernel/etc. i.e. the lower level stuff the program uses to interface with the hardware and o.s. features. The compiler can handle e.g. translation of standard calls to executable code, but it can't magically make a program make different API calls. e.g. you can't make a Qt program suddenly talk GTK or an X program talk Wayland.
I hope that explains a bit better.

[Mike_Walsh]

Off-topic:

Off-topic:

...or do I have that wrong?

You have it wrong.
The issue isn't compiling from source itself, but rather how different distros handle their system libraries/drivers/modules/kernel/etc. i.e. the lower level stuff the program uses to interface with the hardware and o.s. features. The compiler can handle e.g. translation of standard calls to executable code, but it can't magically make a program make different API calls. e.g. you can't make a Qt program suddenly talk GTK or an X program talk Wayland.
I hope that explains a bit better.

Very good explanation. Yes, I see what you mean. Thanks for clarifying!

(And of course, with Windows or MacOS there's only ever the one way of doing everything, so everybody is always "singing from the same hymn sheet", as it were. Which must have its advantages. Hmm...)


Mike.

[Walter Dnes]

Off-topic:

The issue isn't compiling from source itself, but rather how different distros handle their system libraries/drivers/modules/kernel/etc. i.e. the lower level stuff the program uses to interface with the hardware and o.s. features.

It doesn't have to be that low level. When I moved from Gentoo to Devuan linux earlier this year, I tarred up /home/waltdnes/ (minus the .ssh dir) on the Gentoo machine and pushed it over to the Devuan machine. I had a familiar home directory. What could possibly go wrong?

I use mc (Midnight Commander) for file management/rendering/etc. It was constantly falling flat on its face with "file not found" errors on the Devuan machine. A bit of spelunking showed that...
* on the Gentoo machine various utility scripts reside in /usr/libexec/mc/ext.d/
* on the Devuan machine those utility scripts reside in /usr/lib/mc/ext.d/

The Gentoo and Devuan versions of ~/.config/mc/mc.ext.ini have different pathnames. I made the global pathname change with vim and it started working properly.

[frostknight]

But I do firmly believe that there are way too many distros out there, one for every dev who wants an OS to be catered 100% and no less to their own workflow, give or take. Freedom of choice is working against Linux's potential.

Off-topic:
Ubuntu and Debian forks definitely exist too much as in having too many of them.

I also see a divide between people who want to have this personal catering for themselves, and people who have a conviction that they want to force their way of working on everyone else. In the latter group also fall the commercial variants of Linux that are IMHO way too top-down oriented, resulting in part in the reactionary former group who would rather have a distro used by 100 people and being on their own than to bend to that commercial edge.

Off-topic:
Yeah, I thank redhat for being that way. /sarcasm

I dislike them so much.... they are just so damn greedy and manipulative