Pale Moon crashes when I try to log in to Yahoo
 

I installed Pale Moon-27.4.2 in LFS 8.1 and, on the whole, I'm very pleased with it. It looks like Firefox used to look and it runs very fast. But there is one annoying thing it does: every time I try to log in to Yahoo, it crashes with a segmentation fault immediately after sending my name.

I can log in here. I can log into ODF Authors. I can even log into Google, which is a complicated site with loads of scripts and ads, just like Yahoo. But not into Yahoo itself.

It can't be an extension problem because I am not using any extensions.

It's not the end of the world because I have FF on Crux and on Debian (and on my old LFS for that matter) and I can read my Yahoo mail from there. But it's annoying all the same. If there is anyone here who uses Pale Moon and logs into Yahoo, I'd like to compare notes.

I use Pale Moon 27.4.2 on Crux 3.3 and I have no problem to log in Yahoo. As I use it for testing, I didn't bother compiling it, I used the installer for linux. I discovered there is a Crux Pkgfile for Pale Moon 27.4.0 in this repository...

So far I like Pale Moon as well.

You are paying the price using a binary. Build it locally and it won't crash.

Had no luck with building it either. I think I may have to bite the bullet and build FF after all with all the ghastly train of dependencies it carries. Or just use graphical links. There are a few sites though that won't work with that, so my new LFS would be a crippled system. But then admittedly I never put libreoffice on LFS either and that never bothered me; I use Crux or Debian when I need LO.

This morning palemoon crashed repeatedly trying to get into LQ. Had no problems yesterday. I don't think I want software that behaves so unreliably, even though I love the user interface.

Just before the crash, I always get this message:
ATTENTION: default value of option force_s3tc_enable overridden by environment.

A quick google showed that this is a warning, not an error, so it can't be the actual cause of the crashes. But the warning and the crash must have a common cause as they always occur together. The warning comes from javascript apparently, and I noticed this morning that it happened on LQ while advertisements were downloading. Of course once I'm logged in, I don't get any ads, so the browser becomes stable. I bet the Yahoo problem is also due to javascript. That language is a curse on the web!

Have you tried starting the browser from the terminal in hopes that it will throw a useful error message?

It's a long shot, because, when an application crashes, it often doesn't get a change to throw an error, but it might be worth a shot.

Let me have your .mozconfig and I'll build a package for you. Just for fun, to see if it works better. My glibc is 2.25-r5, if yours is older then it won't work.

Yes, of course! That was the first thing I thought of. And the post above your one was the result. Where do you think that message came from?

@emerson: My glibc is 2.26. I'm attaching the mozconfig which is basically the recommended LFS one. The build crashed while compiling the battery manager, whatever that is. And I don't even use a battery in this machine, it's on the mains. Unfortunately there doesn't seem to be an option for disabling it.

OK, I did build and package, the result is (temporary link removed). I noticed there was no -march option in your mozconfig, does LFS not take advantage of CPU specific optimizations?

Standard LFS doesn't. I gather from the mailing list that more experienced users use optimised builds, especially for gmp, but I don't know enough about hardware to do such a thing. I always do a vanilla build.

Thanks for your tarball. I'm going to shut down now and reinstall.

You do not need any hardware knowledge, GCC since version 4.<something> does it for you. Just use -march=native. Of course, you cannot build for another host with native - unless it has the same CPU. You can see what -march=native turns on with

I just noticed this:
Everything was linked to my versions of those libraries. I'm starting to wonder if this was a mistake.

Oops! Your binary needs libjpeg.so.62, which I don't have. I tried to fool it by putting
in a symlink to libjpeg.so.8, but that doesn't work because it actually checks the
version internally. All the other libraries link in correctly.

/usr/lib/libjpeg.so.62: version `LIBJPEG_6.2' not found (required by /opt/palemoon/libxul.so)

The odd thing is that when I googled it, all the references I found were for a 32-bit version. I'm posting this using links.

Hmh ... I'll attempt without system jpeg ... will see ... Having fun with this one.

Alright, I replaced the tarball with new build, jpeg is bundled. I wonder if anything else needs to be bundled.