Remix.run Logo
WhyNotHugo 11 hours ago

Personally, I find the distinction between “issues” and “discussions” annoying.

For one, it duplicates the efforts in checking for prior reports. I might try 5–6 sets of keywords, but now I have to do so for 2 separate trackers.

Tickets cannot be moved between trackers, so instead folks resort to duplicating it and moving discussions… which is entirely opaque if you’re following up via email: you won’t get any more notifications and your future replies are silently discarded.

As a maintainer, having two trackers per project never made sense to me, so I’ve disabled discussion everywhere.

This is mostly a criticism of how GitHub implemented this feature, not of the decision taken here.

emil-lp 9 hours ago | parent | next [-]

> I find the distinction between “issues” and “discussions” annoying

The benefit is that all users who just ask for help, assistance, or are unable to install or use the software now have a place to ask.

You shouldn't create an issue just because you get an error when installing, but it might be beneficial to still ask for help.

If it is indeed a bug, then create a ticket, linking to the discussion.

Normally, too many issues are user errors.

kccqzy 4 hours ago | parent [-]

I personally don’t like discussions either. I treat issues to be issues experienced by the user. It doesn’t mean that there’s an issue caused by bugs in the software. User creates an issue that turns out to be caused by user not following instructions? Just help the user and close the issue.

xixixao 11 hours ago | parent | prev | next [-]

At the very least I think what matters here is the process. The same exact process could be implemented via, for example, issue labels. It would not be hard for maintainers to search for issues with label "bug", which only maintainers can assign. There are clear UX tradeoffs between the two approaches.

vinckr 4 hours ago | parent | prev | next [-]

> Tickets cannot be moved between trackers

You can convert an issue to a discussion and vice versa, so no duplication is needed and your notification should be preserved.

Or do you mean something else?

spinningarrow 11 hours ago | parent | prev | next [-]

Why would checking be duplicated? One would need to check only the discussions in this case, since issues will be created from discussions once something is ready to be worked on (as I understood it)

layer8 10 hours ago | parent [-]

That’s only for non-maintainer submissions. When a maintainer notices a bug or decides on a new feature, they can open an issue for that right away, without prior discussion.

JuniperMesos 4 hours ago | parent | prev | next [-]

I've been annoyed at this distinction myself when trying to troubleshoot problems with other Github-hosted software. I actually think the distinction Ghostty is making here where the Discussion section is open for anyone to bring up an issue with a project, and the Issues section is locked down to official maintainers and is only used for tracking well-scoped planned changes to the software (which might arise from a user-initiated discussion about a problem) makes a great deal of sense. I haven't seen any other Github-hosted project which organizes itself in exactly this way but I have zero problem with it as a user of Ghostty myself.

arccy 10 hours ago | parent | prev [-]

behold the best way to search github issues. Using Google or whatever decent search engine:

    site:https://github.com/org/repo key words
botanical76 10 hours ago | parent [-]

Does this actually work? I would think it will be out of date by at least weeks, and I'd be surprised if their crawlers actually iterate through every issue page.

arccy 9 hours ago | parent | next [-]

At least for the fairly popular repos I work on, it's only a few hours out of date, at most 1 day.

So if I'm triaging a new issue, often it'll show up in the results as well

Jolter 7 hours ago | parent | prev [-]

Google checks my site’s site map at least every day. I rarely see anything out of date in their index.