I probably have the best understanding of the thumbnailing system
among volunteers at the moment, but I'm not interested in bailing out
the WMF on this one.
The fundamental purpose of the WMF is to support the continued
operation of the wikis by maintaining technical infrastructure. There
is not a single Wikimedia wiki that does not use Thumbor.
Nevertheless, maintenance was largely left to a single maintainer
working on Thumbor as a side project, in addition to their normal
responsibilities. That was a decision made or supported by WMF
management. When that single maintainer left the Foundation, there was
no one else to maintain the system.
Sure, I or other volunteers could get appropriate access and finish
off the Thumbor python3 and Debian migrations. But that just kicks the
can down the road instead of actually solving the problem. Maintenance
of a critical system would still be dependent on one person (or more
optimistically, a few people) working in their spare time with no plan
for continuity. That just puts us back to the position we were in a
year ago, which we now know is not sustainable.
Critical infrastructure not directly supported by a WMF team, with a
plan for continued maintenance if a core maintainer leaves, has been a
well-documented risk since before the WMF was created. Yet this is not
the first example of WMF management deferring maintenance from
production systems. In the case of Wikimedia Maps, a full production
outage was necessary before WMF management decided Maps was worth
maintaining. DPL echos a similar tale. What else has to fail before
WMF management starts taking this seriously?
ACN
On Tue, Jan 11, 2022 at 8:31 AM Inductiveload <inductiveload(a)gmail.com> wrote:
>
> On Tue, 11 Jan 2022 at 12:17, Chico Venancio <chicocvenancio(a)gmail.com> wrote:
> >
> > I do not think our goal is to get it done **by paid WMF staff**, but it is also
true that today that is the only viable alternative to get major technical work done. I do
not think it is entirely fair to state that "status quo can be changed by just about
anyone who is motivated to do so (...) just by doing the work". It is not a lack of
motivation that hinders our movement technically, but a lack of resources and shared
governance.
>
> As someone who has tried to contribute using some of my "Covid time",
> and who will shortly be going back to real work, it's also deeply
> frustrating if you cannot meaningfully contribute incrementally, but
> run into a lack of interest in engaging with issues. If someone has
> come along with even a half-decent bug report, it's possible that that
> represents about an hour or so of work: detecting the bug, isolating
> it, creating a Phab account, writing the report all take time. An hour
> of time is actually a lot for someone who commutes to a 9 to 5, let
> alone someone with family.
>
> Now, if that person wants to _fix_ the bug, it took me, very
> roughly[1], three working days to get to a position where I could even
> run Mediawiki locally with the right extensions in Docker[2], figure
> out the code, find out enough about PHP to make a change, actually
> make the changes, and propose a patch for a simple issue, and respond
> to the inevitable CI issues because you can't get the tests to work
> locally. If I had been doing that in the time between getting home
> from the office and making dinner, that's a month of work, and would
> replace all my other interests into the bargain. It's _also_
> frustrating when you then get the patch -1'd because you tried to
> follow existing code, but that's actually old and busted and you
> should use the new hotness that everyone inside the engineering team
> might know, but is highly non-obvious to a person who only cloned the
> repo on Monday. Every review response or rebase also takes a good
> chunk of time, especially if the patch has "gone cold". And then
> there's a good chance your it just rots on Gerrit anyway until you
> have moved on with your life.
>
> I think (making stuff up alert) a reason people want the WMF to help
> here is not just because they collect more than the GDP of Nauru[3]
> per year on the implication that they it supports the wikis, but also
> because they actually employ people specifically because they _do_
> know how to do this stuff effectively, and what is a month of
> sacrificed hobbies for one poor sap is half an hour for them. That's a
> hell of a force multiplier if there was just slack in the engineering
> to allow it. Teach a man to fish and tomorrow he might help you repair
> your jetty. Expect a man to figure out how to mine iron for fish hooks
> on his own and reverse engineer a fishing rod, and he'll be hungry for
> a good while and your jetty is going to stay broken if he wanders off
> in search of something other than fish to eat.
>
> Now, sure, you can say "well it's not the WMF's fault that you have a
> job/family/commute/other hobbies/a herd of depressed pet llamas in
> constant need of hugs, is it?" and yes, you are right, they don't
> officially _owe_ anyone anything. However, it's also a shame to
> completely write off the ability of llama-owners to contribute, unless
> they're willing to put in more time or effort than very many people
> have to spare. Honestly, if I hadn't had Covid time, I would have
> given up on any patch after the first several hours and just walked
> away[4]. And that would have been a shame for me (I'll let others
> decide if it would be a shame for the software!)
>
> So I guess the tl;dr here is that actually a relatively small amount
> of care[5] from the WMF can multiply the effectiveness of outside
> contributions greatly. Those contributions, in turn, can magnify the
> ability of the Communities themselves to Get Content Done[6]. And
> that's the aim here, is it not? We're all on the same side!
>
> --IL
>
> [1] I mean, what did time even mean in 2021, anyway?!
> [2] Since then, Docker has gotten better, though I still can't get
> Phan to work with any regularity, and it doesn't really feel like any
> "staff" developers actually use the method described in
> CONTRIBUTING.md to run up MediaWiki very often.
> [3] Yes, really. To be fairrrr, it's a very small island.
> [4] In fact, before Covid, I tried to do some patches and did exactly
> that, more than once.
> [5] Developer Advocacy is a team that exists, but at least in my
> personal experience, I have never actually encountered it, except for
> bug wrangling.
> [6] Perfect example: ProofreadPage is pretty much entirely non-core
> tech and yet has facilitated almost all work on Wikisource for a
> decade. I'm not sure such a system could be written and deployed
> today.
> _______________________________________________
> Wikitech-l mailing list -- wikitech-l(a)lists.wikimedia.org
> To unsubscribe send an email to wikitech-l-leave(a)lists.wikimedia.org
>
https://lists.wikimedia.org/postorius/lists/wikitech-l.lists.wikimedia.org/