On Di  8.Dez'09 at  0:47:23 +0100, Nicolas Bonifas wrote:
> 
> This trivial patch fixes a minor bug in Maximus: the new window 
> size didn't take the border into account. This bug was particularly 
> visible with the "do not cover dock" option turned on.

I applied your patch to 'next', but the status quo is still not optimal.

The issue with the border and +- 1 pixel-fine-tunning is not good, even
after your patch. Try this. Two windows side by side, not overlapping.
Maximumize the window to the right (ok) and then Maximumize the window
to the left, which now will go fullscreen due to the fact
that there was an overlap of a few pixels before (or that the part of the
algorithm responsible to check if there is another window to the right
is thinking that there is no window to block its full maximization 
because it is not taking into account the borders or +- 1 "safety" pixels)

Another test, if you first Maximumize the window to the left and then
the window to the right then it is ok.

So the Maximus code still needs some more thinking about this border issue
(which was a bit tedious to think because the windows can be borderless, have
no resizebar or frametitle etc).

> BTW Carlos, why do you now have two branches in your repository?

It is not new... the idea was that I wanted 'master' to be stable
in the sense that I would not have to rebase it (changing the
commit ids in the process). But I still wanted to have the possibility
of making mistakes, ie fearless testing patches by putting them
in 'next' and checking if something explodes in the next few days.

Just like your patch now, it is already in 'next' and I am already 
running my wmaker with your modifications. But if after some testing
(by me or others) I decide that I don't want it, I can throw it away 
without ever disturbing the stable branch. 

If some people have multiple screens and want to test Ralf's 
patches from today, they are there too.

But so far the 'next' branch hasn't really caught any big 
issue before reaching 'master'. Due to the small number
of people testing, it is not really all that important in
the process. But it is there.


-- 
To unsubscribe, send mail to [email protected].

Reply via email to