Re: Gnome bug 94684

[tb@becket.net (Thomas Bushnell, BSG)]

Christian Marillat <marillat.christian@wanadoo.fr> writes:

> TB> 1) Upstream author didn't change an API, they changed a direct user
> TB>    issue.
> 
> False.

You know, your utter reluctance to do more than write the minimal
possible words causes frequent problems.  

Here's how it's a direct user issue.  A sawfish user customizes
windows using a GUI customization agent, picks an appearance trait,
and then does a "grab" (they click on window).  That copies the
current Class of the window grabbed into the customization, and all
future windows of that Class will have the customization applied.  

This is a *USER* feature, not an API.  No programming is going on, not
even editing text files with obscure hidden customization thingies,
just straightforward use of a straightforward feature.

Then, when the Class on gnome-terminal changes, the customization of
course breaks.  The bug is that there should be a clean upgrade path,
and not just random breakage of customizations.

> TB> 3) He can report the problem to the gnome maintainers and mark the bug
> TB>    forwarded.  
> 
> Apparently you don't understand. Read my lips ((c) G. Bush) I'll *never*
> change the upstream API, I'll *never* ask the upstream author to change
> that.

1) They don't have to change the API, there are *other* methods of
   solving the problem
2) Does your statement mean you will *never* forward wishlist items
   either? 

> If you don't want to change your configuration each time you did a apt-get
> upgrade, then install potato.
> 
> testing/unstable is for real men (tm).

You don't get it.  A user who upgrades from *potato* to the eventually
released *woody* will get all these bugs.  It is good that they can be
caught now, but they don't just bite users of unstable, they bite
users of *stable* at the point the upgrade occurs.

Thomas