Hacker News Reader

Comment by opportune | Hacker News Reader

I understand when writing paid software with support expectations/contracts, sometimes you have to engage with the teeth-pulling exercise that is “does not work”. But why do you in FOSS? Can’t you just close the ticket and say “not descriptive enough” and move on?

View on HN

To do that, you need a healthy brutal self-respecting trait that I think most of us admire.

The rest of us feel bad doing that, especially towards someone trying to use something we built ourselves that holds some measure of our own pride or even self-worth, it feels closer to obligation. So we waste our time trying to help the characters that deserve our help the least, and we learn to develop a resentful version of that trait over a period of decades.

Sounds like a good way for someone without coding skills to help a project, bug/message triage.

But then, a good bug-reporting system should be able to do such filtering?

I'm wondering if detecting abusive communications is/could be part of it, bit of ML seems like it would fit, even just some Bayesian filtering might do.

One could also train an NN to recognise images of the app (or simpler: OCR|grep) and require a screenshot with any submission.

But I guess automated triage mightn't be the best route. It seems very much a delicately balanced people problem.

I don't think you even need to be that brutal. Just have a file on your hard drive with a canned response that instructs the reporter that they need to provide more information, and how to provide it. Every time you run into a weak bug report, copy-paste that response into the issue comments. If the extra info gets provided, great, now it's a useful bug report. If not, you can close the issue, and not feel anywhere near as bad about doing so.

Like anything else, the more you do it the easier it gets. Its not going to be instant but you'll get used to it.

The Syncthing project will often apply this trait and you're right - it's admirable as well as pragmatic.

It's a skill probably worth practicing.

You can and I regularly close tickets for not following the template that says at the very top "tickets not following this template will be closed with no reply".

It's even better if you can get a bot to automatically reply and close it.

I was thinking about how one could automate this when reading.

Not a foolproof way, but add a string in your Issue Templates that is required to be there, e.g. .

If this is not present in the issue, the GitHub action just closes the issue with a message to please fill in the template if they wish to create an issue.

Doesn't catch nearly everything, but should get some and it's easy to set up. Could be interesting to go further with the idea and maybe check if each section contains text or something like this, hmm...

Another, somewhat more underhanded trick: add a string that is required to not be there, e.g. "Please remove this line entirely.", halfway through the template, and mention it nowhere else.

I've seen this in repos, I'm pretty sure there are bots who implement it. I'd guess it catches more than 90% of cases, especially if you put the string at the bottom of the template.

You can. You can do anything, really, including not reading issue reports at all. But you set out to be helpful anyway, so it’s about finding a cutoff where you don’t want to help anymore. As I said, it’s simply harder than ignoring assholes. (Plus you may get a bad rep.)

Honestly sometimes the amount of things that you need to fill out just makes me use other software. I.e. kmail would never remember that I want my email threads expanded (this example is not real, the kmail devs were very helpful with minimal info in this case). Does this really require I hunt down all this info from all around the place? Sometimes I don't even know where to get the info.

What helps immensely (and I saw this in the Geyser MC project), the software can produce a snippet with all info the devs want with one command, and it even exports to pastebin taking out sensitive info. If you paste such a link in their Discord it even makes an overview with syntax highlighting in the chat. That really helps a lot on my (bug reporter) side. And thus on the dev side.

KDE, and it's related apps, is an interesting one. For a long time you could get the version info from a menu item (Help>About or something) in any of the K apps. But then they changed it to give no version info ... then the bug report tool asks up front what version you're using ...

One of the great things with Steam when I started running it on Linux was it's debugging info that gathered details of your system so you didn't have to.

> But then they changed it to give no version info

This is still the case. Every KDE application has the menu entries Help → About $APPNAME and Help → About KDE which both show the relevant version numbers. I'm overwhelmingly certain this feature never went away because I am on a rolling distro and upgraded through pretty much all versions and I figure I would have noticed the absence of these menu entries.

> then the bug report tool asks up front what version you're using

That's incorrect. The menu entry Help → Report Bug… opens a dialog with version information that has a button Launch Bug Report Wizard which produces a link like e.g. `https://bugs.kde.org/enter_bug.cgi?format=guided&product=kon...`. Consequently in the bugtracker, the available information is already filled in.

----

By the way, this post is an example for the bullshit asymmetry principle, and I resent that I had to spend a magnitude more time to correct your misinformation than it took you to produce it. Please be a better netizen.

Ok, I'm a 20y user of KDE, and a massive supporter. I still got the app version from apt (apt-get at the time) and framework version and still submitted bug reports - after registering and installing symbols to get decent backtraces. I even helped some people improve their forum posts by showing them how to get the relevant version numbers.

There was a change, it was unhelpful. There were reasons, it was to do with changes in versioning on plasma frameworks - I could look up the details, but that isn't really important a few years hence.

The jist of my comment is that structures within an application can help us non-programmers to make useful bug submissions; I cited 2 examples from my real life experience.

Hopefully we agree on the basic premise I set out at least?

Yeah this was also the gist of my comment(s).

Depends on the project. If it is governed by bureaucrats and you do that too often, you'll be accused of "not being inclusive".

The modern version of "Boxer the work horse should work harder".

this is extremely frustrating for the users.

there's a game I used to play fairly often before updates simply broke it. like mission items were replaced with random fires floating in water. many users with the same issue reported it, and some like me even provided a save (which was never even downloaded)

all such tickets were closed with "cannot reproduce"

I'm not (their) tester, I don't have time to fully reproduce issues step by step, and I don't have access to a debug build anyway to figure out the bug trigger condition

"does not work" is the best I can say here.

Was the game free? And did the people working on the game contribute a lot of their time to it for free?

If not then I can see the reason for your frustration, however it is not the same as free software being worked on (at least partly) by volunteers receiving the same lack of effort (or in signal's case nastiness) in bug submissions.

paid

Reply on news.ycombinator.com