Page 1 of 2

Current State of Pale Moon for Linux and the Future

Posted: 2019-09-30, 10:59
by New Tobin Paradigm
Alright you Linux people. Time for some updates.

As you may have read in an unrelated post. Travis is alive and ok and he will return eventually and likely slowly over time. So that is a relief. However, it does mean you are stuck with me as interim Linux Team Leader until that happens.

So here is the current state of Pale Moon for Linux..
Linux 32bit: Pale Moon for Linux mainline (generic) builds for 32bit linux systems are good to go as some may have observed in an earlier thread. Our build is now targeting more legacy systems that have at a minimum of the last GTK2 and glibc 2.12 and other system requirements contemporary to the last 10 years or so built on CentOS 6.x with GCC 4.9. This will be maintained until AT LEAST November 30th, 2020 at which time the build environment (CentOS 6) support ends. After which it is very likely that we will be ending production of Linux 32bit builds.

Any potential 32bit builds that could be produced on a newer platform would raise the glibc version too high for the bulk of users who enjoy the 32bit builds we produce. Now.. Let me state for the record that this will NOT affect the buildablity of targeting 32bit in the codebase and despite GCC 4.9's already "dead" status we will NOT be removing support for it either. All reasonable effort will be made to ensure that you as well as system distribution packagers may create 32bit builds of any UXP Application. We just won't be producing one after next year. To this end, I am personally gonna relax my position on helping users roll their own build for 32bit systems. KEEP IN MIND this is still OVER a year away and unless something changes by then this is the plan of what is going to happen.

Current Linux 64bit: Nothing is going to change with this build except it will be specifically designated as a GTK2 build going forward. It will maintain the minimum requirements that it currently has (at least glibc 2.17 etc) and continue to be built on CentOS 7 and be compiled with GCC 7 (maybe 8 later). Oh and if you want a date put on it? Then it is June 30th, 2024.

NOW FOR THE FUN BIT (assuming the linux 32bit people don't have me executed).. The GTK3 build of Pale Moon.

I want your input on what should be done about the GTK3 build of Pale Moon that I plan to produce and add to our offerings. Specifically, you cutting edge and modern linux distro/version users.. What is your current glibc version.. As well as, should it use latest GCC version that produces a good and stable build? Specifically, and assuming a viable environment can be setup, DO you want this to be targeted at current systems or do you want it to target newer generations of systems. Specifically, should I build it on CentOS 7 like the GTK2 build is or should I build it on CentOS 8 which is approx equivalent of Fedora 29 era systems?

You are the ones who are going to be using it.. So what do you think?

(Please don't make this thread a linux 32bit bitch fest..)

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-09-30, 11:35
by Moonchild
New Tobin Paradigm wrote:
2019-09-30, 10:59
It will maintain the minimum requirements that it currently has (at least glibc 1.17 etc) and continue to be built on CentOS 7
Shouldn't that be 2.17?

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-09-30, 11:43
by New Tobin Paradigm
Yes.

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-09-30, 12:50
by EMH_Mark_I
For myself, most of my systems have glibc 2.23 which is around the Xenial release period of Ubuntu (2016.) That will soon change to 2.27 though as I upgrade my systems.

Is there a particular version(s) of GTK3 that is going to be targeted?

I think targeting newer generation of systems for GTK3 support would be nice; if not easier vs current systems. There's a range of different GTK3 versions out there, so focusing on the last (3.24) might work out best if GTK3 support is targeted for future systems. If you target current systems (which I would guess ranges from 2016 to present), that's going to be difficult.

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-09-30, 15:23
by vannilla
In general I think the best approach with GTK3 is to use the most recent version of everything (OS, library, etc.)
I just don't really trust GTK3 to work that well on "old" systems.

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-09-30, 15:45
by New Tobin Paradigm
Well at least GTK3 isn't a moving target anymore since it is done and merely in bugfix mode. But needing a GTK3 version implies needing newer stuff. GTK2 has been busting on the latest distros like Fedora and they aren't apt to fix it beyond critical failures since it pretty much is already dead. So that is one vote for latest possible providing it works. Also it isn't JUST about GTK3 but also what the minimum glibc version will be.. Which for CentOS 8 will be 2.28.

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-09-30, 16:17
by Alexander
I use openSUSE Leap 15.1, which is the latest stable release of openSUSE, which came out in May 2019. It has glibc 2.26, so I'd prefer if the gtk3 build was done on CentOS 7.

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-09-30, 16:24
by mseliger
I have glibc 2.29 installed (Lubuntu with LxQT, version 19.04).

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-09-30, 16:32
by Isengrim
glibc: :I use Linux Mint 19.2, which comes with glibc 2.27. But Mint isn't exactly known for being super-cutting edge, since it trails Ubuntu.

gcc: I'm not partial to any particular version as long as it works.

gtk: I also vote for targetting 3.24.

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-09-30, 16:59
by New Tobin Paradigm
GTK 3 target version is gonna be the last version. That is a given. The whole point of DOING a GTK3 version is to cover systems where GTK2 isn't predominate or is outright bugged.

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-09-30, 18:53
by Nigaikaze
KDE Neon chiming in here with glibc 2.27.

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-09-30, 19:12
by New Tobin Paradigm
Looks like EL 7 and 8 both use GTK3 3.22 (EL 8 may get an update to 3.24 eventually) so the question is.. CentOS 7 with GCC 7 and glibc 2.17 or CentOS 8 with GCC 8 and glibc 2.28 (Which btw should be in use on the major distros if not newer by now. Debian, Ubuntu 18.10, Fedora).

The thing is, I really think there is a case for the GTK3 flavor to need to target NEW Linux distros from the EL 8 level and beyond while GTK2 version deals with stuff from the EL 7 period.

Here is a build I just built on EL 8: http://repository.binaryoutcast.com/artifacts/builds/palemoon/palemoon-28.8.0a1.linux-gtk3-x86_64.tar.bz2
This does NOT have the updater and the minimum glibc is, as I said, 2.28.

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-09-30, 22:09
by RoestVrijStaal
Since I'm using Kubuntu 18.04.3 with KDE (both on a 32-bit and 64-bit machines), I wish I didn't have to care about the GTK versions as long as there is a Qt-port of Pale Moon.

But since there isn't one of the latter and Pale Moon for Linux depends solely on GTK (at least on libgtk2), I kindly recommend to be conservative about versions, because the developers of GTK seem only to care about Linux+SystemD+GNOME.

Strangely enough I'm using 2.56.4 of libglib2.0 on my 32-bit machine, which is far higher than what I've seen from others here despite I'm using a LTS edition???

Also, isn't there a way to migrate the buildserver to Ubuntu / Debian when CentOS6 goes EOL? Ubuntu 18.04 LTS (inclusive the 32-bits edition) has free support till 2023-04.

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-09-30, 22:32
by New Tobin Paradigm
As has been stated before elsewhere... Debian has specific quirks and oddities that make it completely unsuitable for producing generic builds of anything that is expected to run outside of that family. Despite popular belief there is more than what Canonical produces and/or funds out there. Not to mention that if we used Ubuntu 18.04 it would be almost the same from a glibc perspective as using EL 8 (2.27 instead of 2.28) and that is far too new to be acceptable to the very users of the 32bit build as we found out just a few weeks ago.

Hell, 2.17 was too new. THAT was the whole point of having to rediscover and recreate the CentOS 6 environment. In any case, users of the Debian family are already covered by well produced system packages by stevepusser including 32bit last I checked.

That aside, I think you need to re-read my opening post as well the one before this because you seem to have either terribly misunderstood what I said in its entirety, didn't read it at all, or are just 32bit bitching in which case.. Stop it.

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-10-01, 01:56
by Isengrim
Off-topic:
RoestVrijStaal wrote:
2019-09-30, 22:09
Since I'm using Kubuntu 18.04.3 with KDE (both on a 32-bit and 64-bit machines), I wish I didn't have to care about the GTK versions as long as there is a Qt-port of Pale Moon.
IIRC there used to be an incomplete one in Firefox, way back in the day. Yeah it'd be really cool to have another UI toolkit for the browser, but it would be no small task.

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-10-01, 12:27
by fatboy
I would love a GTK3 build of Palemoon, because I use Arch and Solus Gnome and Solus MATE, which probably us the latest versions of GCC and glibc? I also know Manjaro has Palemoon in their repos. Will post gcc and glibc details later

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-10-01, 12:29
by fatboy
New Tobin Paradigm wrote:
2019-09-30, 19:12
The thing is, I really think there is a case for the GTK3 flavor to need to target NEW Linux distros from the EL 8 level and beyond while GTK2 version deals with stuff from the EL 7 period.
I agree with this. What happens if one uses a "latest GTK3 Palemoon build on an older distro? Will Palemoon look weird, be broken?

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-10-01, 12:51
by New Tobin Paradigm
Well the point isn't so much "Latest GTK3 its self" It's that modern distros HAVE GTK3 and GTK2 may be bugged or even dropped. That is the point.

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-10-01, 15:22
by Moonchild
fatboy wrote:
2019-10-01, 12:29
if one uses a "latest GTK3 Palemoon build on an older distro? Will Palemoon look weird, be broken?
It may or may not run. If older distros don't have the minimum requirements satisfied for the libs it links against (due to the more modern build environment) then it won't run.

Re: Current State of Pale Moon for Linux and the Future

Posted: 2019-10-01, 21:16
by Lunokhod
palemoon-28.8.0a1.linux-gtk3-x86_64.tar.bz2 - Works on Artix, I just tried it here and played a Youtube video.
glibc version : 2.29-4
Doesn't work on Devuan Ascii (Debian Stretch - oldstable equivalent) unlike the current regular GTK2 unstable version, which works fine.

Code: Select all

$ palemoon
XPCOMGlueLoad error for file /opt/pale_moon/palemoon/libmozsqlite3.so:
/lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.28' not found (required by /opt/pale_moon/palemoon/libmozsqlite3.so)
Couldn't load XPCOM.
$ aptitude show libc6
Package: libc6                           
Version: 2.24-11+deb9u4
$ aptitude show libc-bin
Package: libc-bin                        
Version: 2.24-11+deb9u4
But the next release, Beowolf (current Debian Buster stable) should have 2.28 libs. There are no backports to update these libs (as it might break the OS), upgrading to Beowolf testing would probably do it.
GTK4 is in the AUR btw.