We need a nit-fixing-bot

[eshellman]

At least half of all contributions to free-programming-books get tripped up by the linter, which is picky about blank spaces, list collation, pdf links and the like. These are things that could be fixed by a bot, which could add commits fixing minor errors. This would cut the administration overhead significantly and would reduce contributor frustration, improving the experience for all involved.

[borgified]

i googled a bit and found this: https://stickler-ci.com/
i think we should be able to integrate it to our repo but first we should test it out and make sure it fits our needs. ill get around to doing this eventually but if anyone would like to give it a shot and see if it does what it says on the tin, please feel free to do so (and report back the results here)

[eshellman]

Interesting!

[borgified]

i still intend to revisit this soon™ but i came across this while looking for something related which i think is relevant so im leaving it here for later https://github.com/bluzi/travis-buddy

[eshellman]

Maybe someone can try out @TravisBuddy on a fork and report back.

[borgified]

i just tried it in my repo. you can check out the behavior here: borgified#2
if you like that, ill create a PR and then we can merge it. squash commits before merging though, no need to pollute the git history with my testing.

[zhongjiajie]

@borgified I don't think it behavior achive our purpose. In your ofrk PR borgified#2 It just tell contributor whether PR failed or not, and where failed.
we want a bot to tell contributor how to pass CI, or even auto change contribute PR to pass CI test.
Am I right @eshellman

[LincePotiguara]

I looked at the tool to see if it is capable of checking what went wrong in a Pull Rrequest (PR)

Here is what I've found:

• Travis-buddy will check the build log and send back just what didn't succeed
• It has some default checking
• It is possible to add custom rules

The PR created by @borgified show the tool in it's "raw" form.
In order to present nicely the test fails it require configuration. In our case, we would need a script specify the rules.
The best aplication I could find is this with this configuration.
I also research the documentation and the only meaningful example I came across was this.

Conclusion

It's quite possible to hack this bot into this project it will just take some work.

[borgified]

the aim of travis buddy is to make the reason for a failed build more accessible. ie you can see the error as a pr comment rather than have to go into travis. thats all it was intended to do. if you look at my pr you can see various attempts at customizing its behavior but ultimately i went back to the default as this was the most reliable mode. Most of the other repos using travis buddy seem to have arrived at the same conclusion. if we want to have a bot to autocorrect prs that fail the lint test then we should either try my first suggestion or abandon the idea altogether in favor for creating a web form that accepts input in a specific format and then when it is submitted it can figure out how to properly create a pr based on the inputs.

…

On Sun, Mar 3, 2019, 12:36 PM LincePotiguara ***@***.*** wrote: I looked at the tool to see if it is capable of checking what went wrong in a Pull Rrequest (PR) Here is what I've found: - Travis-buddy will check the build log and send back just what didn't succeed - It has some default checking - It is possible to add custom rules The PR <borgified#2> created by @borgified <https://github.com/borgified> show the tool in it's "raw" form. In order to present nicely the test fails it require configuration. In our case, we would need a script specify the rules. The best aplication I could find is this <https://github.com/OsProgramadores/op-website-hugo/blob/master/travis/checkparticipantes.go> with this <https://github.com/OsProgramadores/op-website-hugo/blob/master/.travis.yml> configuration. I also research the documentation and the only meaningful example I came across was this <bluzi/travis-buddy-jasmine-tests#2>. Conclusion It's quite possible to hack this bot into this project it will just take some work. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#2672 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AA4sLFwDUqJ26PoundVl-5gAZFdnqlgtks5vTDJAgaJpZM4ROjqb> .

[ShipsWithCannons]

Stickler looks more promising indeed. The features of Travis-buddy seem to align more with what we already have in the already used Travis CI.

Web forms could work, but that sounds like trading problems instead of fixing them. So if anyone has the time and energy to go figure this out, I'd say focus on Stickler for now.

[borgified]

i went ahead and tested sticklerci and i dont think it'll be able to do what we expect it to do. the automatic fixing is restricted to their list of linters, ie. since we have our own linter with our own specific set of rules, it's not gonna be able to automate fixes.

also i had a hard time getting it to work, im not sure if im configuring it wrong or if it just doesnt quite work: borgified#3 i just couldnt get it to fail

[borgified]

our current linter i think is similar to eslint. i've never written a custom linter before but i've recently discovered that they also have the ability to "autofix" issues. it might be worthwhile to piggyback on the existing linter and implement the fixer portion.

our existing linter: https://github.com/vhf/free-programming-books-lint
googling around: https://flexport.engineering/writing-custom-lint-rules-for-your-picky-developers-67732afa1803

if anyone has prev experience you are more than welcome to take a crack at it, if not, i'll get to it eventually...

[bluzi]

Hello everyone,

Sorry to bump this issue after few months of inactivity, but I'm thinking about implementing an autofix feature for TravisBuddy, and I came across this issue and found your use case very interesting.

Would love to hear whether you've figured it out already and if so how?
Also I am not sure yet how this is going to be implemented for TravisBuddy but I'd love to hear what is the interface that most benefits in your use case, if it makes sense.