Re: runaway .xsession-errors vs. temp races: a compromise?
On Sat, Jul 10, 2004 at 01:04:19PM -0500, Branden Robinson wrote:
> On Tue, Jul 06, 2004 at 02:28:33AM -0700, Chris Waters wrote:
> > [...] people are used to the error file getting truncated. [...]
> > I even made a patch for your consideration.
> For the time being, I do not agree [with the proposal]
Fair enough. I think I understand your objections, and more or less
agree for the most part. My major reservation is still the matter of
surprise on the part of existing users, though, so I reserve the right
to make modified versions of the proposal that may be more to your
taste in future.
> Justification #1: Xsession is a conffile and anybody who's annoyed by this
> can apply your patch or a similar one themselves.
True. I'm not actually annoyed, but this is what I did myself. But I
might propose making it more obvious where and how to patch this.
Also, patching Xsession may require complicated merging at update
time, and unfortunately, dpkg does not include easy-to-use merge tools
for conffiles. So I might suggest moving the code to Xsession.d.
> Justification #2: I think programs and libraries -- especially GTK+ and Qt
> and their dependent hordes -- need to be broken of the philosophy that
> .xsession-errors is just another name for /dev/null.
Hmm. I strongly approve of this motivation! Nevertheless, I'm not
sure I agree with making our users helpless pawns in the battle
between your point of view (no matter how correct that point of view)
and the desktop developers' point of view (no matter how wrong and
muddy-headed that point of view).
But I do agree that your arguments are good reason for rejecting my
existing patch, so I'll shut up unless and until I come up with
something better for everyone.
cheers
--
Chris Waters | Pneumonoultra- osis is too long
xtifr@debian.org | microscopicsilico- to fit into a single
or xtifr@speakeasy.net | volcaniconi- standalone haiku
Reply to: