Re: _NET_WM_[GET_|TAKE_|REQUEST_]FOCUS & urgency

[Luke Schierer <lschiere users sf net>]

On Thu, May 26, 2005 at 09:28:26AM -0600, Elijah Newren wrote:
> On 5/26/05, Luke Schierer <lschiere users sf net> wrote:
> 
> > But both of those would require that you react with the possibility
> > in mind that something may need to be changed, perhaps there is some
> > issue that you didn't already know about, as you apparently already
> > knew that DEMANDS_ATTENTION needed work.  Is that too much for me to
> > ask?
> 
> I did respond like that initially, and tried to gather info on what
> was wrong and what needed to be fixed.  A few dozen emails later (and
> when I thought we had found the real problem--a bug in one of the
> implementations of the spec), you merely repeated the same assertion

what I want is quite simply stated and very hard to actually achieve:

I want to not be able to tell what window manager a user is using by
what bugs he reports.  So far that is true of *most* window managers,
but not all.  I know fluxbox users because they report a bug in
switching tabs in gaim that no one else hits, and I have previously
known metacity users because they complain about gaim "stealing
focus."  Now I am afraid that I will know metacity users because they
complain about "loosing windows."  Interestingly, though I did not
remember this until someone mentioned some of the details of kde
here, I used to see bugs about gaim's "raise window on new im" option
raising the window to *just below* the top level.  I haven't seen
such a bug in some time (a month or two) though, so maybe kde changed
something here?  I do not know.

Still, my overall point remains, if I know your users by what bugs
they tell me, you are doing something wrong, if it was on our side,
then users of every WM would be reporting it.


> that the spec itself is broken and needs to be tossed out for
> something which isn't at all defined.  I'm willing to believe we need
> to change things (we've made various adjustments to the spec since it
> was added in order to address previous limitations), but claims that
> its broken without details of why its broken and an incomplete new
> solution is somewhat frustrating: since we supposedly missed some
> cases in the first spec we came up with, and we don't know what those
> cases are, how could we be expected to fix them with a new spec
> without further detail about the problems or at least some details of
> the proposed solution so we could compare how they worked?  Without
> such details, at best we would merely cover the cases we have already
> handled.  My reaction was likely overblown, but it was merely an
> attempt to point out that if you want to make such claims then we
> really do need details about why it is inadequate.

In this case I think DEMANDS_ATTENTION is perhaps the bulk of the
cause, and so I opened a bug on that and was prepared to consider
that segment of the issue closed here.  Still, on my list of things
that we were concerned with, there were other points. such as
USER_TIME being overloaded and insufficient for the task.
Unfortunately I don't seem to have any notes on *why* we (gaim) came
to that conclusion, and I'm trying to get details on that for you all.

>  
> > Lubos Lunak seems to be confirming our reading that this *is*
> > something that an application might be setting.  it thus *is*
> 
> Yeah, looks like I was confused...
> 
> > WM could be unsetting it? Sorry, that's a somewhat sarcastic reply
> > because I've been told now that its too much work for an application
> > to transfer focus from one window to another.
> 
> I totally agree with Lubos here.  Why should an application developer
> have to worry about calling some function to state something about
> focus for each and every new window it opens?  Many and perhaps even
> most app developers just aren't going to do that.  It's a royal enough
> pain to ask them to call a
> this_is_an_unexpected_window_please_dont_focus_it() function for the
> very few cases where it makes sense.

because unless you call something like gtk_window_present, it didn't
focus on other window managers. that's why only metacity users
complained about gaim stealing focus.  So I would tend to say that as
its already required, it isn't going to be a pain.  But then there
are probly gnome programs that have depended on that focus policy
that *would* have to change, and so perhaps it *would* be a pain for
them.

luke

> 
> > Still, after your somewhat thoughtless paragraph that I replied to
> > separately, you have done a very good job of providing the questions
> > that I need to get answers for you for.
> 
> Yeah, I could have handled that better.  Sorry about that.
> 
> 
> Cheers,
> Elijah
>