1 |
On 11/07/2016 00:06, Philip Webb wrote: |
2 |
> The proliferation of pkgs in KDE, Gnome, Perl + other areas |
3 |
> is going to become a problem for Gentoo, |
4 |
> as they will tend to demand more dev attention |
5 |
> & will also add to users' burden in keeping track of what they use + need. |
6 |
|
7 |
I disagree. My burden is maintaining KDE is about the same through major |
8 |
versions 3, 4 and 5. Nowadays I mostly list the 30 or so KDE apps I |
9 |
actually use in a set, and the USE flags for stuff I have and can use go |
10 |
in make.conf. The ebuilds then take care of things and mostly get it |
11 |
right. When I say "mostly", I really mean a big percentage with lots of |
12 |
9's in it[1] |
13 |
|
14 |
It's important to realise that these new packages are not new software, |
15 |
they are existing software broken up into smaller more atomic chunks. |
16 |
|
17 |
Example, in KDE-3 we had packages kde, and kde-*-meta. These were no |
18 |
atomic, they were "bunches of stuff sort-of somewhat related" like |
19 |
games, network and so on. When you break that up into lots of small |
20 |
packages, the burden goes *down*, in much the same way that software |
21 |
becomes easier when you refactor a giant main() with many global vars |
22 |
into many small self contained functions. |
23 |
|
24 |
I have many times observed comments on -dev where kde maintainers bitch |
25 |
loudly about how difficult it is to maintain the large monolithic kde |
26 |
packages of versions <5. All this seems to add up to the opposite of |
27 |
what you are claiming. |
28 |
|
29 |
Alan |
30 |
|
31 |
[1] Be careful not to commit the human problem of remembering the few |
32 |
times the ebuild got it wrong (especially when using ~arch), and not |
33 |
remembering, or not seeing at all, the many many many times it didn't |