I’ve also written up a transcript.
Hell, I’ll upvote you for the transcript alone!
OK, help me out there.
I work at the National Library of Technology. We license RHEL instead of using CentOS, which we definitely could as we never actually needed the support in the past 3 years. I frequently think about giving back more and more directly.
As I see it, corporations won’t give you a shit unless you make them. The government, on the other hard, has a lot of money to spend and frequently wastes it on proprietary crap and does not yet fully trust the free software, mostly because there are no real authorities to say it’s good.
We can cooperate. There is absolutely no reason why can’t we create an international organization that would:
With that in place, I guarantee that government agencies will slowly start requiring at least the “qualified” level and in time even the “responsible” level of certification from their suppliers. It will also help to make free software more credible.
The 5% profits from contracts would go to the actually deployed projects, where every project would get half of a fair portion (to be defined) and the other half will be given to its listed dependencies.
Crazy or not?
EDIT: That’s about 1000 projects receiving $2000 a year from Prague alone if all suppliers used free software and complied.
I’m not at all clear on the details of how government works in this regard, but personally I think this seems like an excellent idea.
Supporting out-of-date language runtimes (or compilers I guess?) is a major PITA. Testing libraries in particular seem to maintain support WAY longer than other libraries since they are depended upon by other libraries which presumably may need to support legacy versions. I have definitely felt this pain.
FWIW, downstream libraries depending on Hypothesis for testing have given me exactly zero shit about my dropping 2.6 support. The most I’ve got has been a bit of “Argh, this is annoying. I totally respect your decision, but can you help me figure out what the best thing to do here so I can support 2.6 and also use Hypothesis is?”, which is completely fine.
Maybe deep down they know supporting EOL versions is asking too much :) They can always pin to an outdated version of the testing library that works for them, so long as there’s not some game-stopper bug in it.
Still, that is nice to hear. It could be the perception of needing to continue support past EOL is a manifestation of personal anxiety.
Well, it’s also a manifestation of consumer need and bizarre life choices from downstream packagers. Library maintainers are normally not in the target audience of people who actually need this.
It’s possible I’d have got more push back if this was something they had a dependency on to actually run.
(FWIW my suggestion is “Don’t run Hypothesis based tests on 2.6” rather than using a legacy version)