1 |
On 20/02/14 11:23, Steev Klimaszewski wrote: |
2 |
> On Thu, 2014-02-20 at 03:59 -0500, Alexandre Rostovtsev wrote: |
3 |
>> And this is an example of why everyone on the gnome team doesn't like |
4 |
>> the "gtk3" flag. Because well-meaning developers will be looking at |
5 |
>> their one corner of the portage tree, deciding that they are going to |
6 |
>> handle the choice of gtk version without slotting, and not consider the |
7 |
>> effect on the distro as a whole. |
8 |
>> |
9 |
>> You know what's going to happen now, after the QA team decision? |
10 |
>> |
11 |
>> First of all, lots of developers will start renaming "gtk" to "gtk3" in |
12 |
>> their ebuilds' IUSE. |
13 |
>> |
14 |
>> Which means "gtk gtk3" will soon have to be added to USE in |
15 |
>> targets/desktop/gnome/make.defaults (currently, the gnome profile |
16 |
>> globally only has USE="gtk" because the "gtk3" flag is evil). |
17 |
>> |
18 |
>> And users of non-gnome profiles who use gnome applications will of |
19 |
>> course manually add "gtk gtk3" to USE in their local make.conf. |
20 |
>> |
21 |
>> Unfortunately, at the same time, lots of other developers are going to |
22 |
>> start adding support for building against gtk2 XOR gtk3. Because of |
23 |
>> course "Gentoo is about choice", and the more choices, the merrier, and |
24 |
>> the gtk3 flag has been declared as supported by the QA team. And that |
25 |
>> means lots of REQUIRED_USE="^^ ( gtk gtk3 )". |
26 |
>> |
27 |
>> For the gnome team this results in a headache: maintaining a big list of |
28 |
>> "-gtk" / "-gtk3" entries in targets/desktop/gnome/package.use so that |
29 |
>> gnome users get a sensible choice and don't need to edit /etc/portage/* |
30 |
>> just to emerge widely used desktop tools. |
31 |
>> |
32 |
>> But for non-gnome users who manually added USE=gtk3 to make.conf, this |
33 |
>> means regular emerge conflicts after sync, forcing them to *guess* |
34 |
>> whether "-gtk" or "-gtk3" in pacakge.use is the better choice. Maybe |
35 |
>> with portage auto-suggesting the wrong solution just to add to the |
36 |
>> wonderful user experience :/ |
37 |
>> |
38 |
> See, now this is an example of a good email as to why supporting both |
39 |
> can be a hassle for more than just one desktop. Instead of telling me |
40 |
> that I'm dumb for thinking it's a good idea to follow upstream's |
41 |
> supported ideas, and that we should force one or the other. |
42 |
> |
43 |
> The KDE team seems to be able to deal with it just fine, but somehow |
44 |
> it's impossible and hard for the GNOME team. Why is that? What does |
45 |
> KDE do differently that makes it feasible? |
46 |
> |
47 |
> |
48 |
|
49 |
No, they didn't manage it, at all, which why we don't see Qt3/KDE3 in |
50 |
tree anymore. |