weston does not fallback gracefully if $XDG_RUNTIME_DIR is not set

Preferably don't add this dependency. This should be seeded as part of ubuntu-desktop, and weston itself should fall back to $something_sensible if the path isn't set.

from the spec http://standards.freedesktop.org/basedir-spec/basedir-spec-latest.html
"""
If $XDG_RUNTIME_DIR is not set applications should fall back to a replacement directory with similar capabilities and print a warning message. Applications should use this directory for communication and synchronization purposes and should not place larger files in it, since it might reside in runtime memory and cannot necessarily be swapped out to disk.
"""

I am not sure what good fallback location is.

And that is why there isn't one.

"[libpam-xdg-support] should be seeded as part of ubuntu-desktop...."

Is that a personal opinion, or something that is actually going to happen?

> Is that a personal opinion

No, this is the whole reason pam-xdg-support was written in the first place.

Individual applications should not, in general, recommend specific pam modules.

> I am not sure what good fallback location is.

Check the behavior of e.g., nautilus or dconf for comparison. I believe they fall back to something under the user's homedir - that's not guaranteed to be local, but it's guaranteed to exist and be writable.

Great, thanks. In that case I'm fine with weston having no recommends for libpam-xdg-support.

From #weston:

--- Day changed Thu Sep 06 2012
15:52 < smallfoot-> weston should work without XDG_RUNTIME_DIR
15:52 < smallfoot-> as per the spec
15:52 < smallfoot-> http://standards.freedesktop.org/basedir-spec/basedir-spec-latest.html#variables
15:52 < smallfoot-> "If $XDG_RUNTIME_DIR is not set applications should fall back to a replacement directory with similar
capabilities and print a warning message. "
16:13 < Darxus> smallfoot-: That's kind of funny.
16:20 < krh> smallfoot-: the funny thing is, there is no directory with similar capabilities
16:33 < smallfoot-> krh, oh, maybe ~/tmp/ or /tmp/ or /run/user/ ?
16:34 < smallfoot-> today mountall was updated in ubuntu from 2.38 to 2.40 which have /run/user/ and the preliminary work for XDG_RUNTIME_SUPPORT, the other part is missing is a PAM module
16:36 < smallfoot-> https://bugs.launchpad.net/ubuntu/+source/consolekit/+bug/894391
16:38 < thiago> $HOME could be network-mounted
16:38 < thiago> it could also be SMB, which means no sockets
16:38 < thiago> /tmp is insecure, since other users can write to it
16:39 < thiago> and /run/user/$USER *is* XDG_RUNTIME_DIR. If your distro creates it, why doesn't it set the variable?
16:43 < smallfoot-> oh
16:43 < thiago> but we could create a secure dir in /tmp
17:27 < uartie> smallfoot: http://lists.freedesktop.org/archives/wayland-devel/2012-June/003935.html
17:28 < Darxus> Heh.
17:29 < Darxus> uartie: There were problems with people not realizing it was using ./ and as a result trying to run clients from a different directory than where they ran weston.
17:30 < Darxus> Creating a secure directory in /tmp seems good.
17:30 < uartie> yeh

The link in there is to a post by krh (creator of wayland) to the mailing list on the same subject: http://lists.freedesktop.org/archives/wayland-devel/2012-June/003935.html

I don't really mind weston just failing if $XDG_RUNTIME_DIR is not set. I think it cut down on confusion considerably.

My interest in this bug is resolved with ubuntu-desktop now having a dependency on libpam-xdg-support.

Is there any real chance anybody here is going to do anything on what the summary was changed to, the lack of fallback if $XDG_RUNTIME_DIR is not set, especially now that it will be set by default for everyone due to the above, or should it be closed "won't fix"?