1. 25
  1. 33

    I empathize with accidentally making your repo private and “losing” your community, but this read more like a rant blaming GH because the OP made the mistake of clicking through a confirmation menu; that makes it hard to have any sympathy.

    1. 28

      It’s not (just) a rant, it’s an ad. Gotta get those k’s of star back somehow!

      1. 7

        What’s ironic is that I probably was one of those stars, but the edgy jabs at MSFT/GH (short of going /. on them with that EEE link drop) convinced me I probably don’t care about their progress anymore as the part of the product I use is feature-complete enough

      2. 3

        Yeah, I don’t particularly disagree with the points about how the design could be improved, and how maybe GitHub could be a little more accommodating about restoring it, but at no point there is an actual, honest, no but-they-also-screwed-up-so-it-is-not-really-my-fault, acknowledgement of the fact that the author fucked up.

      3. 14

        The title alone tells me we put way to much stock on trying to make source code ‘social’.

        I asked a junior recently to survey out in the world a type of tool we needed. Every tool he came back with was JavaScript or TypeScript, and when I asked why, he gave a response about it being popular on GitHub. While JS isn’t inherently bad, it ended up topping all of ‘charts’ because of the JS familiarity bias. It also only led dev to only consider the options on GitHub. Nothing he did was wrong, but I’m very much concerned with this mentality that stars have a lot of actual value—that not technical merits or code quality but popularity are so important that the author made a 9–11 minute read post about it.

        1. 1

          too* much stock (too late for the edit)

        2. 9

          idk, stars != community, but this still sucks, it’s easy to have empathy here

          What started as a side project has recently become a company and our team is growing HTTPie into an API development platform

          sus radar blips

          1. 6

            Starring a project subscribes you to a subset of notifications, e.g. about new releases, so loss of stars does have effect on the users, not just on the maintainer’s ego. Of course, stars is a horrible mechanism to subscribe to new releases, but that’s another story.

            1. 2

              yep! i totally understand. i only meant to point out that losing stars doesn’t mean losing the whole community that has been built up over the years. active maintainers will keep being active, bug reporters will keep reporting bugs, and so on.

              1. 2

                Starring doesn’t opt you into notifications. I’ve started 1000s of repos and I’d be flooded with notifications if it did. You only get notifications from watching the repo, they also lost all watchers making it private.

            2. 8

              Ok. So if GitHub does whatever database heroics needed for this user, where does it end? 10k users or more? 1K? Ten?

              GitHub operates at enormous scale. I don’t know how hard it would be to roll this back, but I’m guessing it’s non-trivial. Doing it once? Ok. But imagine all the “oops, we screwed something up, can you restore it for me?” requests that would follow?

              And if this is the attitude this user takes when the fault is entirely theirs, wait when GitHub turns down the next user because they weren’t a big enough star on GitHub.

              Ask yourself, it you were responsible for handling these exception requests, if you’d want to be doing it if everyone acted this entitled.

              1. 2

                The tweet they link in the article has this response. Apparently they’ve done it for some folks in the past but it caused more work than it was worth by the sounds of it.

                1. 4

                  That dialog contradicts what the article suggests. The author of the article would have had to type httpie/httpie which should have raised some big giant red flags, you’d think.

                  1. 6

                    I agree it’s mostly a rant. However, there’s a real UI issue there: the “profile README” repo conventions for users and organizations are different. For users it’s github.com/$username/$username (e.g. dmbaturin/dmbaturin). For organizations, it’s github.com/$organizationName/.github. Cognitive inertia can be really dangerous in this situation.

                    I don’t think there’s an excuse for not remembering that httpie/httpie is the name of your actual repository when asked, but I think his UI improvement suggestion is actually good: confirmation dialogs for destructive actions should display the scale of the consequences.

                    1. 3

                      I definitely agree with you. I’m just pointing out that (I don’t remember) him pointing out that he had to type httpie/httpie and that seems suspect.

                      1. 3

                        No, he admits it in the post that he typed httpie/httpie out of cognitive inertia from the way “profile readme” repos work for user accounts.

                        I didn’t realize at the moment there’s an inconsistency in the naming of this special repo containing profile READMEs and that it differs for users and organizations: name/name vs. name/.github.
                        That’s why I proceeded to make httpie/httpie private instead of httpie/.github without realizing my mistake.
                        1. 2

                          Well crap. My bad then!

              2. 4

                While GitHub’s UI/UX can be at blame, it’s ultimately a user error. When you have incidents where the root cause is an accident, it took a person to cause this one.

                Many actions on a computer are akin to walking around with a loaded firearm and a hair trigger. Things that delete the entire filesystem tree (i.e. executing random binaries as superuser; old-school ruses to delete System32; etc) are examples of how destructive unintentional user actions can be.

                When GitHub instructs you to type in the full name of the repository (sans .git), it’s akin to the other officer with the other nuclear launch code, while you have instructions to kill him and install his code otherwise.

                TL;DR: don’t shoot yourself in the foot.

                1. 2

                  you didn’t lose your community I hope, otherwise they would be pretty superficial just for a counter. your software is still valuable without your pride. I don’t get this whole idea of “github deleting a community” and “killing 55000 people” (wtf), this vocabulary seems totally overkill to me. I agree that losing push notifications is important tho, bit it should mechanically be far less than 54k watchers.