[Slackbuilds-users] Situation regarding dbus and webkit
dev at iotib.net
Thu Sep 29 22:12:21 UTC 2022
On 9/29/22 17:23, Dave W. wrote:
>>> Hi all,
>>> So there's a situation upstream with webkit, dbus and the epiphany browser.
>>> Basically, in future versions of webkit, because the default dbus socket
>>> path is an abstract socket, it will crash pretty much no matter what.
>>> What needs to change to get around this, is to change
>>> <listen>unix:tmpdir=/tmp</listen> in /usr/share/dbus-1/session.conf to
>>> Now upstream dbus is considering making the latter default. Now my question
>>> for all of you, is simply. What do? When I next update epiphany should I
>>> include instructions for the user to modify their session.conf to get it
>>> Relevant issues:
>> Have you read /etc/dbus-1/session.conf and tried making an e.g.
>> session-local.conf or files matching session.d/*.conf ?
> Sorry, that was incomplete - should have said that if that works,
> then make a .new file for doinst.sh to install.
> sbo-maintainer-tools (sbolint, sbopkglint):
> SBo templates:
> How to format github downloads:
While this did work, now I'm getting another issue where webkit is
crashing xdg-desktop-portal, ugh. And I'm a bit of a loss on how to fix
that. I mean, it works, and I haven't noticed anything go unstable from
xdg-desktop-portal being dead. But I dunno.
Seems the workaround is to set the environment variable Malloc=1 before
running webkit. Apparently it forces webkit to use the system's malloc.
While this still crashes xdg-desktop-portal, it doesn't crash webkit. I
don't know the security/stability implications of using Malloc=1. But
nothing seems to break even with xdg-desktop-portal crashing still.
Also, since the dbus thing is a webkit problem. Should the .conf file
for the non-abstract dbus socket go to the webkit package?
More information about the SlackBuilds-users