> We should try to come up with a solution that alleviates the burden on
> the maintainers. Given how often this issue arises, what if we try, as
> a collective, to suggest new mechanisms that would improve the
> situation?
IMO one thing that would help is to somehow losen the current very tight
coupling (the guix monorepo).
a high level of centralization always translates to a bottleneck, and guix has
reached a point where its centralized structure is seriously hindering it and
frustrating away contributors.
i.e. it's much less risk to delegate the management of a hypothethical python
channel to someone trusted than to give commit access to every contributor who
wants to help maintaining the python related packages.
so, i think a lot of packages should be in channels. probably everything that
is not essential for a minimally functional system that can bootstrap itself.
part of python could be in the main guix repo, but whatever is not tightly
needed could go into a channel with its own access control management.
some channels would have loser standards (god forbid some wouldn't frustrate
the contributors with the ChangeLog format in the commit messages), and the
users could decide whom they trust with what. and the guix maintainers could
decide which channels to list on the official web page, with what
comments/warnings.
but this is easier said than done, because formally expressing the dependencies
between these channels is not a trivial task (think of e.g. keeping `guix
time-machine` working in an environment where a patch in the official guix repo
can break a package in some random channel, which is then fixed there in a new
commit, etc). and even if doable, it would probably introduce extra code
complexity.
but i'm sure there were countless discussions about the pros and cons of the
monorepo, and i wish such a topic was captured in a wiki where the main
arguments, obstacles, and ideas were collected for people like me who are late
to the party.
but then guix doesn't really have an official wiki either. (and no, an
afterthought in the libreplanet wiki doesn't count. and if you have an urge to
argue, then just look at its current contents...) so we're pretty much stuck
with the flat mailing list archive and IRC logs.
hrm, a tangential: maybe digesting the guix mailing list and IRC archive will
be my first actual use-case for an LLM... modern problems require modern
solutions.
--
• attila lendvai
• PGP: 963F 5D5F 45C7 DFCD 0A39
--
“Millions of people never analyze themselves. Mentally they are mechanical
products of the factory of their environment, preoccupied with breakfast,
lunch, and dinner, working and sleeping, and going here and there to be
entertained. They don't know what or why they are seeking, nor why they never
realize complete happiness and lasting satisfaction. By evading self-analysis,
people go on being robots, conditioned by their environment. True self-analysis
is the greatest art of progress.”
— Paramahansa Yogananda (1893–1952)
