On Fri, Aug 24, 2007 at 10:01:16AM +0200, Georg Baum wrote:
> > And those resources would not have been available anymore for other
> > work.
> 
> Do I really have to tell you that not wasted resources are not automatically
> available for something else?

No. I know that's not always the case.

> [...]
> And of course you are the one to decide who wastes his resources and for
> what, and other developers are not entitled to decide themselves whether
> they want to be "nice" and do silly stuff or not. If you are really serious
> with that:
> 
> - Why did you not remove gtk at the same time as qt3?

One reason was that qt4 was supposed to be complete replacement to qt3
and I never really tested Gtk, so I was not sure I was not missing
something important. Another (and probably more important) reason was
that the Gtk frontend did not attract core developers, so it did not eat
the kind of resources I was concerned about.

> - Why did you not work towards one build system far earlier,

Because I did not to too much work on LyX in the last few years so the
build system creep did not really hit me. Apart from that, bot cmake and
scons are more or less invisible if they are not used, and both Peter
and Bo never complained when their respective toy was broken by someone.

The current activity is mainly triggered by my feeling that LyX building 
is so slow that it hampers development. I know for sure that each time 
_I_ tried to do some 'real' LyX work in the last few years I got
distracted by build issues quicker than the work itself was ripe.

>       and why did you create a fourth one?

One reason is that I do not want a to start a shoot-out before all
potential contender had a chance to show their nice sides. A second
reason is that qmake has two abilities than none of the other three
systems have: It knows about Qt, so no special code for .ui or such
needed, and it has _real_ MSVS integration including project management,
integrated help, designer and so on based on .pro files as primary
source of information. It was also the easiest way for me to produce
dynamic libraries which I needed to test my hypothesis that using
static liking is part of the current build slowness.

Also, the final solution in the build system arena might as well be some
hybrid solution as e.g. using autoconf for the actual configuration and
using the fastest 'build backend'. 

> - Why don't you stop the windows installer nonsense?

Because I already wasted half a year of my lifetime on Windows installers
and I won't touch them even with a three yard pole unless there was a
_really good reason_. [And I doubt your purse is big enough to qualify
as "_really good reason_" ;-}]

> The two versions are wasting far more resouces than qt3 could have
> ever wasted, because there is no date in sight for the merging, and
> because far more people are involved.

You may be right. I ignored that battle so far. Would you think it
helped if I just threw a dice and removed one of them [I am rather 
more than less serious here]?
 
> And the rest of you should better get a consensus on the goals for 1.6
> rather soon and then stick to that (even those who do not fully
> support each goal). 

We started a list, didn't we. What does it look like now?

Andre'

Reply via email to