1 |
On 9 February 2016 at 01:46, Michał Górny <mgorny@g.o> wrote: |
2 |
> 1. It is a conflict-induced fork. As such, it will never be merged |
3 |
> upstream and it will never be supported upstream. In fact, it is |
4 |
> continually forces to follow upstream changes and adapt to them. eudev |
5 |
> is more likely to break because of the Gentoo developer(s) working hard |
6 |
> to merge upstream changes to their incompatible code. |
7 |
> |
8 |
> 2. Many of Gentoo users are programmers who appreciate the 'vanilla' |
9 |
> API experience Gentoo often provides. Switching the defaults to a fork |
10 |
> that is known to intentionally diverge from upstream goes against that |
11 |
> principle. Programs written against eudev may not work correctly with |
12 |
> upstream udev. |
13 |
> |
14 |
> 3. eudev has fallen behind systemd/udev more than once in the past, |
15 |
> and caused visible breakage to users this way. |
16 |
> |
17 |
> 4. eudev is underdocumented, and the maintainer admits that 'he sucks |
18 |
> at documenting'. In fact, did anyone even bother to note how far eudev |
19 |
> diverges from upstream udev to this point? |
20 |
|
21 |
|
22 |
These problems can be resolved, but the resources involved with |
23 |
resolving them are not trivial. |
24 |
|
25 |
For instance, all it would take to change #1 and #2 would be for |
26 |
`eudev` to become an external project, not an implied child of Gentoo, |
27 |
and for our tools to consider it as such. |
28 |
|
29 |
That is, instead of users being encouraged to see eudev as "Gentoos |
30 |
Udev Fork", they'd have to see it as a stand-alone fork of udev with |
31 |
the goals of retaining the axioms and values of Unix systems and |
32 |
simplicity. |
33 |
|
34 |
Then it would be resorted to the question of "Which external competing |
35 |
udev implementation makes sense for our default value in conjunction |
36 |
with our other default value of rc systems: OpenRC" |
37 |
|
38 |
#3 and #4 are the more important criticisms IME, and we need eudev to |
39 |
be the responsibility of more than one person and have a |
40 |
survival,reliability and progression strategy that will continue in |
41 |
the event upstream drop udev entirely to the point that Gentoo cannot |
42 |
simply port their changes anymore. |
43 |
|
44 |
Because if *that* happens, the fragmentation will become a much bigger |
45 |
problem, and systemd getting bug fixes and updates while eudev gets no |
46 |
love is not really much of a long term solution. |
47 |
|
48 |
In short, for eudev to be a viable long term solution, we might need |
49 |
to have a guarantee of investing a lot more resources in the project |
50 |
than we presently can afford. |
51 |
|
52 |
And until we have some kind of tacit assurance that we can do this, |
53 |
making it the default might not be the best choice. |
54 |
|
55 |
That said, if the upstream fragmentation is going to go in the "no |
56 |
backports and udev goes away" direction, users currently taking the |
57 |
"udev" choice are in for some bumpy waters eventually anyway. |
58 |
|
59 |
|
60 |
-- |
61 |
Kent |
62 |
|
63 |
KENTNL - https://metacpan.org/author/KENTNL |