1 |
On Thu, Sep 10, 2015 at 7:31 AM, Vadim A. Misbakh-Soloviov <mva@×××.name> |
2 |
wrote: |
3 |
|
4 |
> I disagree witho you and hasufell. |
5 |
> |
6 |
> It *IS* users destiny if they get some stabiity issues because of their |
7 |
> decision to have gtk2-only or gtk3-only system. |
8 |
> |
9 |
> Yes, they can paste bugs about improper toolkit support. Is it bad? Rules |
10 |
> says |
11 |
> it should be reported upstream. And all the time Gentoo exists that worked |
12 |
> this way. |
13 |
> |
14 |
|
15 |
I agree, and yet, I wholly disagree. |
16 |
|
17 |
In the before times, the maintainers were often users. They maintained |
18 |
packages and added support for non-standard configurations precisely |
19 |
because they needed these configurations; they wanted them and they |
20 |
experimented with them. |
21 |
|
22 |
As the distro grew in size, in userbase, and in package count, you see this |
23 |
experimentation shunted off into other areas. |
24 |
|
25 |
1) Local overlays. Often if users need to do a thing, they can simply use |
26 |
epatch_user or similar local over-rides to gain the functionality they |
27 |
desire. Gentoo itself has a fair amount of tooling to make this easy; its |
28 |
still way easier than doing it in other distros (the oft-mentioned Debian |
29 |
for example...) |
30 |
|
31 |
2) Published overlays. We have overlays, and layman, and often discovering |
32 |
that someone else has added support for the customizations you want is |
33 |
fairly straightforward. Of course, it could be easier. Of course, we could |
34 |
put all ebuilds in one giant repository. Unfortunately with users comes |
35 |
some standard of reliability. In the early days I don't think anyone |
36 |
equated Gentoo with reliability. I think there is some higher standard now |
37 |
as many organizations have built atop Gentoo and expect some level of |
38 |
reliability (now whether this is a sane expectation is a separate |
39 |
discussion..but I digress.) |
40 |
|
41 |
|
42 |
> |
43 |
> The whole point of Gentoo is to give user freedom of the choice. Freedom to |
44 |
> decide every aspect that is possible to decide about. Freedom to use Gentoo |
45 |
> exactly as they want, but not as "you don't need feature X, because I'm |
46 |
> maintainer/QA and said that", like some DebUbuntu maintainers did with git |
47 |
> or, |
48 |
> say, ejabberd, some years ago. Any movements to the easy side of "we will |
49 |
> not |
50 |
> support feature X, despite upstream still support it, because feature Y is |
51 |
> newer and shiny, and feature X can be less tested" is a big fat violation |
52 |
> of |
53 |
> Gentoo philosophy. |
54 |
> |
55 |
|
56 |
I absolutely refuse to allow this user-choice to be used as a stick to beat |
57 |
maintainers into doing whatever users want. The maintainers are the ones |
58 |
doing the work, and they get to choose. Many maintainers are sympathetic to |
59 |
user choice (as you note, it is a component of the distro philosophy) and |
60 |
many maintainers go out of their way to support user choice. But it is not |
61 |
a cudgel. |
62 |
|
63 |
|
64 |
> |
65 |
> And I totally agree with Rich: it is maintainer decision, if they ready to |
66 |
> support mutiple build variants or not. And if not — it is absolutelly |
67 |
> lawful |
68 |
> user's right to file a bug against a package, that it has support in |
69 |
> upstream, |
70 |
> but has not in the Gentoo. |
71 |
> |
72 |
|
73 |
And its absolutely OK for a maintainer to close the bug as WONTFIX after a |
74 |
lively discussion. |
75 |
|
76 |
|
77 |
> |
78 |
> WE HAVE NO RIGHT TO DICTATE users what they should use and what they should |
79 |
> not. We are makers of kinda army swiss knife suite that give user |
80 |
> possibility |
81 |
> and instruments to make everything they want. And any tries to say "you |
82 |
> shall |
83 |
> use SystemD, but not sysV/openrc/upstart/whatever", or "you shall use gtk3 |
84 |
> only", or "you shall use Qt5 only", and so on — is a CRIME against Gentoo |
85 |
> philosophy. |
86 |
> |
87 |
|
88 |
> -- |
89 |
> Best regards, |
90 |
> mva |
91 |
> |