1 |
Jason Stubbs wrote: |
2 |
> |
3 |
> |
4 |
> There's ways to manage this complexity, such as putting the dependencies into |
5 |
> autotools' RDEPEND (if it can be considered correct) or by using |
6 |
> meta-packages. However, your point is against requiring that packages _must_ |
7 |
> specify all system dependencies. While I personally believe that packages |
8 |
> should specify all dependencies, what I'm arguing against is requiring that |
9 |
> packages _must not_ specify any system dependencies. |
10 |
> |
11 |
> -- |
12 |
> Jason Stubbs |
13 |
|
14 |
I agree with your personal belief, however I also find it unmaintainable |
15 |
in the current system (metapackages in their current form |
16 |
non-withstanding as I don't think they are a great solution, merely duct |
17 |
tape if you will, but that is another discussion entirely). |
18 |
|
19 |
There is no benefit for me as a package maintainer to dep on a system |
20 |
package unless there is an existing problem. From a maintainer POV it's |
21 |
extra work and extra writing to keep the deps up to date. Also there is |
22 |
the whole thought of what to list? Do I list only glibc versions that I |
23 |
know work? gcc versions that I know compile my code? Where does the |
24 |
line get drawn? What is the point of depending on certain elements if |
25 |
say, they are already a dependency of $PACKAGE_MANAGER? It is not |
26 |
pragmatic for a distribution to do so IMHO, 'technically correct' or not. |
27 |
|
28 |
A few use cases that would go against this involve people doing odd |
29 |
things like emerging a bunch of stuff and then unmerging a critical |
30 |
system package (say, ncurses); since my program happened to depend on |
31 |
ncurses anyway it would 'fix' the problem automatically for the user |
32 |
instead of dying with no ncurses. Of course this use case could be |
33 |
handled another way; by having portage make sure that packages listed in |
34 |
'system' are installed. While I am a fan of the 'fix it automatically |
35 |
in DEPEND' way here; the use case is rather...convenient. Unmerging |
36 |
many things in system either break portage, or won't affect anything (oh |
37 |
no I unmerged virtual/editor!) |
38 |
|
39 |
So yeah, in conclusion; too much work, fix it when it's reported broken |
40 |
seems like a decent (pragmatic) policy to me. If/When it becomes easy |
41 |
to list stuff (package sets or something, please not metapackages in |
42 |
their current form) then I'd be much more interested in implementing it. |
43 |
|
44 |
As an aside, If you are unsure in a given situation feel free to ask |
45 |
someone about it. Worse case you (put an extra dep in|leave out a dep); |
46 |
both are easily repairable. |
47 |
-- |
48 |
gentoo-dev@g.o mailing list |