1 |
On 11/22/2014 11:20 PM, Rich Freeman wrote: |
2 |
> On Sat, Nov 22, 2014 at 1:54 PM, hasufell <hasufell@g.o> wrote: |
3 |
>> |
4 |
>> No one would care in such a distributed model if there is one person |
5 |
>> blocking progress somewhere. They would just move on, regroup around a |
6 |
>> new overlay and start working there and let that guy/project rot forever. |
7 |
>> |
8 |
> |
9 |
> Nobody can block progress under the current model. If you feel |
10 |
> otherwise, please point them out so that they can be dealt with. |
11 |
> |
12 |
|
13 |
They can block progress and they do. And by saying we allow conflicting |
14 |
ideas in one repository we are even making it worse. |
15 |
|
16 |
The council is a workaround to make the broken project structure not |
17 |
look too bad. |
18 |
|
19 |
>> |
20 |
>> We don't need more authority, we need less... and we need more actual |
21 |
>> opensource workflow. Our tools, our organizational model and our |
22 |
>> workflow are ALL ancient. And they don't seem to work very well, do they? |
23 |
> |
24 |
> Gentoo is already fairly non-authoritative where the main tree is |
25 |
> concerned. I'm all for more overlay support, but I doubt it is going |
26 |
> to fix the kinds of issues you're bringing up. |
27 |
> |
28 |
> The problem with java is that nobody wants to work on it. Lots of |
29 |
> people want to talk about working on it, but nobody is writing |
30 |
> ebuilds. |
31 |
> |
32 |
> The problem with games is that nobody wants to work on those either. |
33 |
> Lots of people like to talk about the games project blocking progress, |
34 |
> but now that this has been eliminated, there isn't some flood of new |
35 |
> games ebuilds. |
36 |
> |
37 |
|
38 |
I strongly disagree. I know a fair amount of games overlays where people |
39 |
do work on games ebuilds. They just don't give a sh*t anymore to try to |
40 |
get that stuff into the main tree, because they were alienated long ago. |
41 |
|
42 |
The image of the games team is so bad, that not even gentoo devs bother |
43 |
anymore (except me, uh). Yet neither the council, nor comrel has done |
44 |
anything radical, except giving recommendations, asking for them to |
45 |
elect a new lead, blah blah. |
46 |
|
47 |
In a distributed model this project would just have been abandoned by |
48 |
the community 8 years ago and people would have started a new fresh |
49 |
overlay. Currently this all sucks, because it will conflict with in-tree |
50 |
ebuilds and because we don't have good enough tools for this kind of model. |
51 |
|
52 |
> People love to talk about elitist old-timers blocking progress, but it |
53 |
> seems to me that many of the old-timers don't do a whole lot of |
54 |
> anything. I think the complaint is really that other people aren't |
55 |
> doing the work we want them to do. |
56 |
> |
57 |
|
58 |
It's not about elitist old-timers, it's about a more dynamic model that |
59 |
has low tolerance for |
60 |
* bugs being open since 8+ years, because no one bothers to |
61 |
review/change stuff (check nethack bug) |
62 |
* territorial behaviour |
63 |
* slacking devs slacking so hard, but not stepping down |
64 |
|
65 |
In addition, this model requires a workflow that is long overdue, |
66 |
including proper VCS like git or mercurial and a review culture. None of |
67 |
this happens on a larger scale. Instead we are stuck with tools like |
68 |
bugzilla for ebuild reviews and push our happy ebuilds to the CVS |
69 |
repository. |
70 |
|
71 |
So now guess again why people don't bother, because: |
72 |
* have to become gentoo devs over a period of 6 months or so, then |
73 |
realize they are stuck with territorial crap, people ignoring each other |
74 |
and have to appeal to the council, comrel or whoever multiple times |
75 |
before something happens? |
76 |
* or they have to write bugs reports on bugzilla, attach ebuilds |
77 |
manually, get a partly review in a timeframe of 9 months if they are |
78 |
lucky, re-push attachments, start again |
79 |
* or they can try to contribute to sunrise which may be simirlarly slow |
80 |
(mind you, I've been a sunrise dev, so we can talk about that if you like) |
81 |
* or they just start their own overlay and stop caring to collaborate |
82 |
with gentoo devs |
83 |
* If they are very lucky, then their favorite project already uses an |
84 |
overlay-workflow (e.g. haskell, science). And those projects usually are |
85 |
so slow with moving their overlay ebuilds into the tree, that it's |
86 |
almost useless doing so. They should just stop and focus on their overlays. |