You must log in or register to comment.
I’ve been there, but over the years I’ve gotten better at avoiding being in this situation.
If you are implementing something for yourself, and merging it back upstream is just a bonus, then by all means jump straight to implementing.
However, it’s emotionally draining to implement something and arrive at something you’re proud of only to have it ignored. So do that legwork upfront. File a feature request, open a discussion, join their dev chat - whatever it is, make sure what you want to do is valued and will be welcomed into the project before you start on it. They might even nudge you in a direction that you hadn’t considered before you started.
Be a responsible dev and communicate before you do the work.
Somebody please fork reprepro, there’s a super useful bugfix in one and a super useful feature in the other but I want both.
The bugfix is the zstd decompression-cancel race condition bug and the feature is multiple versions per package but they’re both super stale.
Maybe…
Maybe I can fork it…Fork the feature one, get the diff of the commit that patches the bug and apply the diff to your fork.
Now compile and test.
I don’t have a lot of experience with merging but then again, this is a great opportunity to learn.
This is why I always ask if the maintainers are open to a PR first.
I once made a PR changing
a = f(b)froma = "" if b == "" else f(b). About six months later still inpending review. I get that people have their own shit but damn.edit) Sorry. Clarified the code.
It seems like you are implying this is an obvious optimization, regardless of context. Why do we not care what’s going on with
b?Yeah, the two aren’t equivalent and the original has more conditions than the new one, so without context this just doesn’t make sense in this example.
A is “” only when B is also “”, otherwise we return f()
In the new one we simply say that regardless of what B is, we’ll just call f(), entirely skipping the case where B == “”.
Probably this specific condition checking was moved to the inner scope of f(), but this example does not tell us (who don’t know the context) that. Or maybe the check is redundant, but that also isn’t signaled in any way.
Or then maybe I’m just oblivious to the optimization, in which case I can see why the maintainer would take their time figuring that out. It’s not anything obvious based on that alone, at least to me, and I would say I have some experience in this field.
Edit: But yeah this is basically just semantics, I’m sure they gave apt description in the PR, so the context would be explained there and none of this really matters. I just like to ruminate about little things like this for some reason. Didn’t mean to imply they didn’t do a good PR, just that this specific example was either confusing or confused.
I’ve missed the
f(b)part. Edited.
Or worse, ban you from their GitHub repository without any reply or explanation
“closed by stalebot”
Even better when someone makes the exact same PR and it gets merged a few days after being opened and yours left unreviewed.
The repo owner is lost in the mountains and will accept your request after they get out of the hospital and find a new job and apartment in two years. :)
If that’s happening to you that’s crazy. GitHub is way too noisy though. I get 30 notifications on that apps notification widget though for just bullshit I didn’t even know I signed up for or snyk or some other garbage.
I did this, then it too so long that it broke. Still hasn’t been pulled.
And this is why I don’t contribute. Or at least I’ll ask a question about whether or not something would be a desired feature and if I don’t get a clear yes or no by someone who can actually approve a PR, I. ain’t. coding. shit.
This is how its supposed to be done anyway.
Fair enough, but as someone who has worked closely with the Decky Loader maintainers and contributed my own stand alone plugin I get it. We basically all have day jobs as devs and it can be mentally taxing to do more PRs at home. Not to mention sometimes there’s just not enough time in the day, and I don’t even have kids.
Maintainers are ultimately volunteers doing work with hundreds of dollars an hour for free. I’ve had some PRs take 20+ days to be looked at, it’s just how it goes.
You’re framing this as if it were something unusual. Unsolicited PRs are a no-go in my opinion. It’s just basic communication and collaboration to align with the maintainers whether a change is actually required or not.
That’s the fucking worst, when you put all this work into a free and open project only for the lead to be like “nah don’t like it”
Free and Open as in, free to do the work for them and Open for it to be rejected for almost no reason.
A project is in no way, shape, or form obliged to to accept and maintain your code, especially if it’s not a feature they want. If you want your feature so badly, maintain a soft-fork yourself. Don’t want to put in that effort yourself? Then why should the project maintainers?
