[spec] Is the process for updating the spec working?

Jonathan M Davis newsgroup.d at jmdavisprog.com
Fri Jul 5 05:56:31 UTC 2019


On Thursday, July 4, 2019 5:58:17 PM MDT Dibyendu Majumdar via Digitalmars-d 
wrote:
> I have a small update that is languishing for weeks.
>
> I also see other updates that are pending.
>
> It puzzles me why this is the case.
>
> It seems to me there should be one person responsible for
> accepting changes and applying them, and not have to wait for
> random reviews and comments. So who owns the spec, and why isn't
> the owner stepping up?

That's not really how things tend to work around here (maybe it should be,
but it isn't). There are code owners for modules in druntime and Phobos in
the sense that it's set up so that github will bug certain people to review
stuff in those modules based on who's worked on them, but for most modules,
there isn't a true owner. Nicholas Wilson has been paid to act as the review
manage for pull requests to try and keep things moving along, but other
people really need to be reviewing stuff in addition to him, which doesn't
always happen as often as it should.

As for the spec, I don't know that we even have it set up to notify specific
people about PRs for particular pages like we do with modules in druntime
and Phboos. Certainly, there is no one person who "owns" the spec. There's
also the issue of how much Walter needs to be involved with spec changes.
Something that clearly improves the spec without really changing what it's
saying should be reviewable by a lot of different people, but anything that
starts actually changing the semantics could require Walter - or at least, I
don't think that it's clear when it would or wouldn't. And if Walter's
approval is required that could become a bottleneck. In general though, I
suspect that you're just suffering from the general problem we have in that
we aren't getting enough people reviewing PRs, so too many PRs languish in
the queue, and the spec is something that a lot fewer people are going to be
willing to mess with, so it probably suffers even more than the other
projects.

- Jonathan M Davis





More information about the Digitalmars-d mailing list