Discussion:
Segfault with development dillo
Jeremy Henty
2014-04-14 06:22:35 UTC
Permalink
This URL reliably segfaults Dillo:

http://freethoughtblogs.com/butterfliesandwheels/2014/04/where-will-brandeis-go-to-get-its-respect-and-honor-back/

I suspect this is due to a very recent change as for the past few days
Dillo has been crashing a lot, which until then hadn't happened in
ages. I haven't investigated further.

Dillo (current hg tip), Linux 3.13.9, glibc 2.13, gcc 4.5.2 .

Regards,

Jeremy Henty
Sebastian Geerken
2014-04-14 13:46:21 UTC
Permalink
Post by Jeremy Henty
http://freethoughtblogs.com/butterfliesandwheels/2014/04/where-will-brandeis-go-to-get-its-respect-and-honor-back/
I suspect this is due to a very recent change as for the past few days
Dillo has been crashing a lot, which until then hadn't happened in
ages. I haven't investigated further.
Dillo (current hg tip), Linux 3.13.9, glibc 2.13, gcc 4.5.2 .
Yes, everything after revision 3613 segfault; you can stick to this if
you want to run the development version. (There are no important
changes yet after this.)

This is part of a couple of changes which I hoped to solve another
problem, but which did not serve this purpose, so I'll probably soon
revert these changes again.

Sebastian
Jorge Arellano Cid
2014-04-14 14:30:33 UTC
Permalink
Hi Jeremy,
Post by Jeremy Henty
http://freethoughtblogs.com/butterfliesandwheels/2014/04/where-will-brandeis-go-to-get-its-respect-and-honor-back/
I suspect this is due to a very recent change as for the past few days
Dillo has been crashing a lot, which until then hadn't happened in
ages. I haven't investigated further.
Dillo (current hg tip), Linux 3.13.9, glibc 2.13, gcc 4.5.2 .
Yes, after the dillo-3.0.4 release, the large work on floating
elements was merged into the repository, leaving a very unstable
and crashy dillo.

Users should stick to dillo-3.0.4 in the meantime.

IMO, we (developers) need to focus as a team on stabilizing
this tree before planning any further enhancements. Only once
this goal is met we can return to our usual workflow.
--
Cheers
Jorge.-
Jeremy Henty
2014-04-14 19:33:15 UTC
Permalink
IMO, we (developers) need to focus as a team on stabilizing this
tree before planning any further enhancements. Only once this goal
is met we can return to our usual workflow.
Completely agree.

It's just like old times! <nostalgia/> :-)

Maybe I can update my wrappers to automatically generate tracebacks
from core files. Hmm...

Regards,

Jeremy Henty

Jeremy Henty
2014-04-14 19:26:25 UTC
Permalink
Post by Jeremy Henty
http://freethoughtblogs.com/butterfliesandwheels/2014/04/where-will-brandeis-go-to-get-its-respect-and-honor-back/
In fact as far as I can tell *any* page from freethoughtblogs.com
crashes Dillo, which suggests that it is site style that is causing it
rather than the content. The page content downloads without any
problem and Dillo starts rendering it and downloading images for a few
seconds, then crashes. This suggests to me that it is still
downloading the CSS and crashes when it tries to style the page. (I
run Dillo inside valgrind so everything happens around 30x slower for
me. I can clearly see Dillo restyling the page once the CSS has
downloaded.)

Regards,

Jeremy Henty
Loading...