Re: gnome-terminal awfully slow

[Bill Haneman <bill haneman sun com>]

On Thu, 2003-02-13 at 00:00, Havoc Pennington wrote:
> On Wed, Feb 12, 2003 at 10:49:46PM +0000, Bill Haneman wrote: 
> > The fact that vte was compiled into 2.2 was *not* highlighted in the
> > runup to 2.2.  I think it's quite possible that our terminal
> > accessibility has considerably regressed from 2.0, and that's not
> > good.
> 

Sorry for sounding so annoyed about this, but I still think it was pooly
resolved.

> My memory is that we had several long threads about this which you
> were involved in.

Yes, but the end decision/action was not called out as it should have
been, i.e. there was lots of back-and-forth in which differing opinions
were aired.  The actual change should still be explicitly noted,
*especially* where there has been controversy in the past.

> 
> I think I made it quite clear that unless someone fixed Zvt, I was
> moving to VTE. Accessibility is important, but we aren't going to
> stick to seriously broken/inadequate/unmaintained code just because it
> is equally broken for everyone.

My impression was that the most recent discussions of VTE leading up to
feature freeze noted a number of significant performance issues; it
certainly did not look to me as if there was consensus to make the
switch for 2.2 (though many people did advocate that course of action).

I think this is one instance of an ongoing problem with GNOME, in that
consensus is not always reached and the actual end decision (which
usually falls to the maintainer) is not, in the end, telegraphed
appropriately. i.e. 

objector: "The issues I raised are still open"
maintainer: "I told you what I was planning to do, why should I tell you
after I've done it"

IMO in the above case (where there are clear open issues), the
maintainer should always explicitly let the concerned parties know that
the change under discussion has actually taken place.

I think that if we are going to say anything publicly about GNOME
accessibility (and we have already), we must avoid replacing code that's
broken in one way with code that's broken or regressed
accessibility-wise.  Regression is not generally tolerated in other
areas, it should not be tolerated in accessibility either.

-Bill

> Havoc
-- 
Bill Haneman <bill haneman sun com>