News for hackers. Stuff that matters.
Re:Dear GitHub by ggreer on Thursday January 14, 2016 @06:00PM (#10905471)
I maintain a C repo[1] and user idiocy is much lower than what I've seen in JS projects of similar popularity. Still, I agree with these criticisms of GitHub. I hate +1 spam enough to delete such comments. Sometimes I even ban those who do it. I'm frustrated by people who open idiotic issues[2][3][4][5]. I procrastinate on bad pull requests because my options are:

1. Close the PR with little or no comment. People then think I'm an asshole.

2. Spend hours explaining why the code is terrible and why it can't be improved. In addition to being a big time sink, PR submitters often don't understand the criticisms. Half the time, they still think I'm wrong.

People even defend stuff as obviously wrong as adding a thousand lines of GPL'd code to an Apache-licensed project.[6] Then they say I should remove .gitignore support from ag because it doesn't implement 100% of .gitignore syntax. As if users would be happier with tons of extraneous results instead of some extraneous results.

A lot of this is cultural, but GitHub could help steer things in a better direction with the features proposed in this letter. I hope they take this letter seriously.

1. https://github.com/ggreer/the_silver_searcher

2. User accuses ag of hard-locking his computer: https://github.com/ggreer/the_silver_searcher/issues/791

3. User wants ag to always print filenames, unlike every other tool out there: https://github.com/ggreer/the_silver_searcher/issues/749

4. User wants ag to replace PCRE with a totally different, incompatible regex library: https://github.com/ggreer/the_silver_searcher/issues/698

5. User aliases 'ag' to 'grep', then complains ag doesn't work: https://github.com/ggreer/the_silver_searcher/issues/578

6. https://github.com/ggreer/the_silver_searcher/pull/614

Re:Dear GitHub by monkmartinez on Thursday January 14, 2016 @07:13PM (#10905979)
I actually hesitate to really "watch" repos because the sheer volume of email generated is staggering. I take my hat off to those who run successful OSS projects. Thank you!
Re:Dear GitHub by imtringued on Thursday January 14, 2016 @06:33PM (#10905676)
About number 5. I don't see anyone complaining about anything.
Re:Dear GitHub by ggreer on Thursday January 14, 2016 @06:43PM (#10905752)
Maybe "complain" is too strong, but he created a GitHub issue –notifying several hundred people– without running ag --version. Heck, he didn't even look at the output of his command. It was immediately obvious to me, from the limited information he provided, that it was a bash alias.
Re:Dear GitHub by chris_wot on Thursday January 14, 2016 @07:19PM (#10906006)
Looks like a genuine mistake. He apologised. You've never done something dumb and didn't realise it?
Re:Dear GitHub by timv on Thursday January 14, 2016 @07:26PM (#10906050)
I think this gets to the root of the GH problem

To the person raising the issue it's a simple mistake, sorry.

To the person that has to deal with it, it's yet another issue being raised where the reporter didn't follow the necessary steps to diagnose the problem themselves and (implicitly) expected a bunch of other people to apply their own time to solving it.

If the "New Issue" form had a place where the reporter was asked to paste the output from ag --version then it might have caught the accident before it wasted the developers' time.

I think it's an over reaction to describe the issue as a complaint, but it is an example of how the GitHub UI forces project admins to deal with incomplete and poorly investigated issues from users.

Re:Dear GitHub by alextgordon on Friday January 15, 2016 @10:41AM (#10909899)
All successful products will have stupid users. If you don't want to have stupid users, don't write successful software.
Re:Dear GitHub by ggreer on Friday January 15, 2016 @12:41AM (#10907452)
A mistake, I can understand. But each issue I linked to was not a mistake. Each one was the end result of a series of mistakes stemming from a combination of ignorance, negligence, and (occasionally) incompetence.

Have I done dumb things without realizing? Of course.[1] But in almost 20 years of software development, I have never created issues resembling the ones I linked to. Bug reports are seen by hundreds of people and take up valuable developer time, so I make sure mine are useful.

To use an analogy: Say I'm giving a talk to an audience of a hundred people. I wouldn't do it extemporaneously, without slides, then walk away in the middle of Q&A. And if I did, I wouldn't call it a mistake. I'd call it being a terrible presenter. Yet that's what bad bug reports are like:

User (notifying hundreds of people): "It doesn't work."

Dev: "What version are you using? What error messages do you see? How are you running it?"

User: * crickets *

It's gotten bad enough that I wrote a short post on how to report bugs.[2]

1. https://news.ycombinator.com/item?id=7145451

2. http://geoff.greer.fm/2015/08/15/how-to-write-good-bug-repor...

Re:Dear GitHub by ams6110 on Friday January 15, 2016 @08:57AM (#10909288)
Why would you even respond to a bug report that was "it doesn't work."

Such a vague report tells you all you need to know about whether it's going to be worth your time trying to work with the person who submitted it. Close it immediately as non-actionable.

Re:Dear GitHub by chris_wot on Friday January 15, 2016 @04:32PM (#10912264)
As much as I respect you for The Silver Surfer, that 5th issue was a very silly mistake.
Re:Dear GitHub by shurcooL on Thursday January 14, 2016 @10:02PM (#10906852)
Do you really want people to be afraid to report an issue in case it's something silly/not an actual problem? Would that be better?

I guess it depends on the project and its contribution guidelines.

Re:Dear GitHub by ggreer on Thursday January 14, 2016 @11:16PM (#10907142)
> Do you really want people to be afraid to report an issue in case it's something silly/not an actual problem? Would that be better?

I seriously doubt the pendulum will swing too far in the opposite direction. Right now, the majority of created issues are close to useless. If those people took five minutes of their own time to troubleshoot, it would save others hours.

Re:Dear GitHub by sondr3 on Thursday January 14, 2016 @06:38PM (#10905709)
True, but it's still user problems that aren't problems with the software itself. Far too often you get people who want help with _everything_ on a project, from installing the right language, their editor and so on when it's not something that really concerns you.
Re:Dear GitHub by chei0aiV on Thursday January 14, 2016 @08:56PM (#10906539)
SourceForge fixed the +1 issue recently.