1 |
On Sat, Jul 18, 2020 at 3:00 AM Alec Warner <antarus@g.o> wrote: |
2 |
> |
3 |
> So I think we can, somewhat, get away from this framing. Just get the community to pitch N projects and rank them. Offhand I can think of a few: |
4 |
> - Replace portage with pkgcore. |
5 |
> - pkgcore maintainer (replace radhermit; who has been asking for help for months) |
6 |
> - build a package-build service that is easy to deploy. |
7 |
> - More performance work on portage. |
8 |
> - Rewrite catalyst. |
9 |
> - Replace repoman with pkgcheck. |
10 |
> - Compile repo format into something less terrible then we have today. |
11 |
> - Replace rsync mirror network with GIT https serving. |
12 |
> - Help arzano build more developer tools on top of p.g.o |
13 |
> |
14 |
|
15 |
Again, not intended as criticism, but as one other thing to think |
16 |
about and maybe add to this list. |
17 |
|
18 |
Some of these changes are purely back-end and don't impact dev workflow much. |
19 |
|
20 |
Some of these changes will or could impact dev workflow. You need to |
21 |
include in these projects the efforts to get the rest of the devs |
22 |
trained/transitioned over to the new way of doing things. |
23 |
|
24 |
Right now big changes don't happen very often, so we don't really need |
25 |
to worry about this stuff. If you start commissioning "more developer |
26 |
tools" or "replace repoman with pkcheck" and so-on, then you need to |
27 |
consider that this change impacts the volunteer devs who aren't |
28 |
getting paid to keep up with all the changes. |
29 |
|
30 |
How many GSOC projects have come out with some new tool and maybe 5% |
31 |
of the devs use it? Often this is because the tool was built, a short |
32 |
article was written on a blog (maybe), and that was the end of it. We |
33 |
already have lots of tools that I bet 80% of devs don't use to their |
34 |
potential. They have a workflow that sort-of works and they don't |
35 |
spend much time on Gentoo, and in general when we build new tools the |
36 |
docs are pretty minimal, or are more focused around reference than |
37 |
workflow. |
38 |
|
39 |
So, if you're going to start commissioning changes you'll want to be |
40 |
careful that the unpaid volunteers don't fall behind. I've seen |
41 |
non-profits do stuff like this and it rapidly segments the org into |
42 |
the 5% of the org that schedules all their meetings M-F 9-5, and the |
43 |
95% unpaid volunteers who become disconnected from all the |
44 |
decision-making and become less involved. The org then backfills |
45 |
their work by hiring more. It initially works because 1 employee |
46 |
probably can do the work of 20 volunteers, but then 95% of the |
47 |
community basically gets relegated to just paying the bills. The |
48 |
crisis happens when the 95% start feeling uninvolved enough that they |
49 |
STOP paying the bills, and then the whole organization collapses. The |
50 |
95% don't stop paying the bills out of resentment - just disinterest. |
51 |
People care more about an org if they actually directly contribute to |
52 |
it. |
53 |
|
54 |
I do think it is important to not lose sight of letting the community |
55 |
directly contribute. It helps keep the leadership grounded in what |
56 |
the community wants, and it keeps the community around. |
57 |
|
58 |
Again, this isn't a reason to never spend money. You just have to |
59 |
make sure that every time you're spending $1 on some new bit of |
60 |
tooling, you're spending another $1 on keeping everybody else involved |
61 |
in it. That isn't wasted money - it is the cost of staying relevant. |
62 |
|
63 |
-- |
64 |
Rich |