1 |
Nikos Chantziaras posted on Sat, 28 Jul 2012 13:07:08 +0300 as excerpted: |
2 |
|
3 |
> On 28/07/12 12:27, Ralph Sennhauser wrote: |
4 |
>> On Sat, 28 Jul 2012 15:54:07 +0800 Ben de Groot <yngwin@g.o> |
5 |
>> wrote: |
6 |
>> |
7 |
>>> We do not have (nor want to support) a qt useflag. We have opted for |
8 |
>>> "qt4" and "qt5" useflags as the most straightforward and least |
9 |
>>> confusing. |
10 |
>> |
11 |
>> Indeed, the flag qt has almost disappeared from the tree. Good to know. |
12 |
> |
13 |
> Why the different policies between the gtk and qt USE flags? This just |
14 |
> looks inconsistent. |
15 |
|
16 |
Different gentoo projects. Different people involved with their own |
17 |
preferences. But I believe it's mostly an accident of history. |
18 |
|
19 |
The gtk/gtk2 evolution went rather poorly as IIRC there really wasn't an |
20 |
original defined policy, so the gtk USE flags were ambiguous. At first |
21 |
USE=gtk2 was discouraged for a lot of packages, since for them it meant |
22 |
favoring the still (at the time) less stable gtk2 over gtk1. USE=gtk |
23 |
meanwhile, sometimes meant favor gtk1, while at other times it meant let |
24 |
the package maintainer pick the best one to support. Of course that |
25 |
caused problems later on, after gtk2 matured and gtk1 was being phased |
26 |
out, so a general policy was adopted, that AFAIK remains today: USE=gtk |
27 |
meant support gtk in any form, with USE=gtk1/gtk2 (and now gtk3, with |
28 |
gtk1 phased out) meant prefer that specific version instead of letting |
29 |
the package maintainer choose a default. |
30 |
|
31 |
But the key point there is that said policy was kind of fallen into by |
32 |
accident, and once in place, it was simply more convenient to maintain |
33 |
it, then to change it yet again. |
34 |
|
35 |
When the qt3/qt4 case came along, they had the lessons of the gtk case to |
36 |
examine and decided to avoid the problem by switching to specific- |
37 |
versioned qtX flags I believe before/as qt4 hit the tree. Of course the |
38 |
fact that the existing in-tree support was already qt3 helped, since that |
39 |
was already more intuitive than gtk1. From quite early on, then, simple |
40 |
qt was never allowed the ambiguity of gtk -- it always meant qt3 but was |
41 |
quickly deprecated in favor of the qt3 flag. |
42 |
|
43 |
Of course also helping things was the fact that the qt3 ecosystem was |
44 |
much more monolithic and kde3 much more dominant within it than was the |
45 |
case with either gtk1/gnome1 or the now somewhat broader-ecosystem qt4/ |
46 |
kde4. So getting buy-in for the quick deprecation of qt in favor of qt3 |
47 |
was much closer to simply getting by-in from the gentoo/kde folks (with a |
48 |
large overlap between them and the gentoo/qt folks), as opposed to the |
49 |
wider cooperation needed in the gtk case. |
50 |
|
51 |
So to a large extent the fact that gtk means any gtk while the versioned |
52 |
ones mean prefer that version, while there's ONLY the versioned qtX |
53 |
flags, is an accident of history. And since then, the respective gtk/qt |
54 |
policies have remained in place due to inertia -- yes there's an |
55 |
inconsistency between them, but users of each quickly get comfortable |
56 |
with it, and the cost-benefit ratio of trying to change either one now, |
57 |
simply hasn't been considered worth it. Thus as new versions appear, |
58 |
gtk3 and now qt5, they simply follow type. |
59 |
|
60 |
-- |
61 |
Duncan - List replies preferred. No HTML msgs. |
62 |
"Every nonfree program has a lord, a master -- |
63 |
and if you use the program, he is your master." Richard Stallman |