Implementing extension methods in a way that does not slow down performance...

[jasnell]

This does two things:

1. If an extension method is detected, an optional callback is called
2. All the characters in the method are validated using TOKEN(ch)

[ghost]

+1 in the sense that I'm interested in this and would like to see it happen downstream in tmm1/http_parser.rb as well.

[dfa1]

+1
I'm testing this patch, it works very well

[rlidwka]

+1

[indutny]

Generally, looks fine. Mind fixing some style issues?

[indutny]

Also, please rebase it over the latest master.

[sintaxi]

Anyone know of the status of this PR? I would love to see node get custom HTTP methods.

[indutny]

@sintaxi have you read my messages above? I think we could land it, but some fixes are required first.

[sintaxi]

@indutny sorry, didn't mean to be dense. I read your messages but also see that the contributor has not been seen in this thread in 5 months and it looks like mostly formatting issues holding this back.

@jasnell are you still active on this PR?

[indutny]

@tjfontaine do you think it'd be fine if I'll land it with fixes myself?

[tjfontaine]

@indutny if the user has signed the CLA you should be fine to land with your own fixes, if you change functionality you should maybe do it in a second commit though

[indutny]

I'm afraid that @jasnell didn't... @jasnell may I ask you to sign http://nodejs.org/cla.html ? It's the only real obstruction on the path of landing your PR.

[jasnell]

Semi-active. I've been doing a few other things here lately. Can return to this later on this week if that's ok.

[indutny]

Sure, thanks for being at least semi-active! :)

[eliasgs]

@jasnell any progress with this?

[chmac]

@jasnell Were you able to sign the contributor agreement? Would be great to see this PR land...

[mk-pmb]

👍

[chmac]

@jasnell Just a quick follow up, any chance you've already signed the Node contributor agreement? I'd love to see custom HTTP verbs supported in Node... :+1

[chmac]

@indutny You still open to fixing the issues in order to merge this? With the recent removal of the CLA it seems like this is ready to move forward... 👍

[indutny]

@chmac the problem with this patch is that we have very simple and short method detection in http-parser.

Basically, right now it thinks that COPY, CAPY, CQPY` is the same method. I don't think that there is much use in this implementation. Sorry.

However, here is a rebased and improved version of this patch: https://gist.github.com/indutny/3403029c5815fd0799ee .

[chmac]

OK, I'm not sure about the implementation, I'm hoping to get custom http verbs in node... :-)

Is the gist ready to be merged? Does it need to be converted into a pull request?

[jamesamcl]

Basically, right now it thinks that COPY, CAPY, CQPY` is the same method.

Can I just say. That's pretty horrible.

[indutny]

@chmac the gist is finished, but please don't open a PR for it, because it is very unlikely that it'll be accepted.

@udp this is the price for speed, I guess. Though, I haven't benchmarked it.

[chmac]

OK, so the gist is not a good solution, this PR is now not a good solution, is there another option?

Had something changed since this PR was ready to cleanup + merge? I spent a bit of time chasing a CLA, then that became unnecessary, but now we can't move forward with it. :-( Custom http verbs felt so close...

[indutny]

Heh, I feel your pain.

Well, perhaps we could introduce some limited support with the exceptions that I have mentioned. It'd be cool to hear @bnoordhuis opinion on it.