Comment by sytelus | Hacker News Reader

Comment by sytelus | Hacker News Reader

Parent Full thread

Linus says:

> I think the trivial patches are among the most important ones - exactly because they are the "entry" patches for every new developer.

I have been running a fairly successful open source project and this is not exactly my experience. I guess about 70% of pull request I get is simple spell, grammar, white spaces and the likes. Some time folks send this in wholesale because they want their personal favorite verbiage. It takes enormous amount of time to go through each line and each file to make sure something else is not sipping in. However, more importantly, rarely I have seen these text fixers to send any real code fixes. Vast majority of them are just one offs.

My thesis is that there are lots of folks out there just trying to brag to be "contributor" for popular open source projects and they just move on from one project to next rarely taking time to settle down in any individual project. This might be not be the case for Linux kernel but might be the case for other smaller projects.

> takes enormous amount of time to go through each line and each file

>My thesis is that there are lots of folks out there just trying to brag to be "contributor" for popular open source projects

Do you have an example of this? All I could find in your top 3 repos was one pull request updating the readme and its only 8 lines.

https://github.com/sytelus/CryptoJS/pull/3/files

To be fair, he works for Microsoft, its entirely possible he maintains one of their OS repos.

A more generous interpretation (and the one that I think explains my very short stints as a contributor in just the way you describe) is that I really appreciate the project, want to help somehow, but won't/can't commit to deep work on the project. I also see many projects that practically beg for help "even if it's just documentation" and believe small contributions to be similar to adding a dollar to the tip jar.

>It takes enormous amount of time to go through each line and each file to make sure something else is not sipping in.

This is an enormous exaggeration.

I've done this for a few projects, and it's almost always a PR to fix a specific pain point that I've had, that isn't obvious from the PR itself. Examples are

- downcasing some http headers in an example (because they wouldn't match unless downcased, which I spent hours figuring out)

- replacing a regex search with a string search (because it made the method 30x faster and it was on the critical path of a weird edge case)

- allowing the user to prevent some harmless automatic stats being loaded in a postgres library on connect (because it broke cockroachdb, which uses the same wire protocol)

- quoting the urls in some generated CSS (because in base64 "//" is sometimes a valid encoding and our CSS preprocessor implemented line comments)

Now, I left a description for all of these that explained the problem, but without context they'd just look like I was collecting "contributor" labels.

The only time I've really seen this has been in October, when DO and GH ran their "Hacktoberfest" promotion. Open 5 PRs during October to get a free T-shirt, and some people really just pick random repositories and send random tiny changes. I really liked Hacktoberfest in the past, but this year these people really soured it to me.

E.g. look at 22-52 days old denied PRs on the HTML spec: https://github.com/whatwg/html/pulls?q=is%3Apr+is%3Aclosed

That's an entirely different thing to "serious" small fixes that actually clean up small problems in docs and such.

Not my experience at all. And even if true, it wouldn't matter. Someone found a problem.

They can

a) ignore it

b) file a bug/shout on irc

c) do a pull request.

I much prefer c over b, and both over a. Every little bit helps, and it's improving the current state. I've hardly seen people doing extensive (as in hard to review) commits that looked like busywork to get an 'I commited' badge of honor (is that a thing?) in any repo.

And I also do that all the time, if I find some thing that's quicker to fix via PR than even find the external bug tracker or the irc channel? Everybody wins.

Reply on news.ycombinator.com