Please provide test vectors for IRCv3.2 message format changes

[kaniini]

The topic says it all, would be nice to have test vectors that are 'officially' valid, and ones which are officially 'invalid' so we can test for conformance in the parser.

[clokep]

This would be useful, using this in a test suite for emitting messages is probably more useful. You'll likely want to ensure you can parse the 'invalid' messages anyway (if you believe in "be strict in what you send, but generous in what you receive", at least).

I've collected some 'conformant' and busted messages over the past few years that I've seen in the wild (or had reported to me in bugs): http://dxr.mozilla.org/comm-central/source/chat/protocols/irc/test/test_ircMessage.js Hopefully it's useful to someone. (In particular I find some busted ones from Unreal interesting.) These pretty much deal with server-client messages, not sure if that's what you're after or not.

[grawity]

I have some at https://github.com/grawity/hacks/tree/master/irc/parsers/tests for testing my own IRC parsers. Though I will need to update them a bit; I've recently realized that test-irc-split is not particularly useful; but some of them might be interesting anyway.

(@clokep What kind of 'unescaping' can create newlines inside a message?)

[clokep]

@grawity Seems to be referring to the CTCP "Low Level Quoting" (that's what CTCP document calls it, at least). I...don't seem to have any tests for how this is handled in the CTCP parsing, and I don't know if I've ever seen a client actually send text like that besides Instantbird/Thunderbird (and I believe we no longer send any messages like that).

[dwfreed]

I've yet to see a client that actually does the quoting and unquoting specified in the CTCP "spec." (ChatZilla, the more common extension for Mozilla software, does not, afaik). It's not a great idea.

[kythyria]

@clokep That document technically permits >1 CTCP message per PRIVMSG. I don't think I've ever seen that happen either.

[clokep]

Yes, it does allow that. I think we all know that the CTCP protocol is not entirely sane. I wrote code that tries to parse weird messages, but as I said we don't send them (back to my "be strict in what you send, but generous in what you receive" tenant).

I've never seen them in the wild that I've noticed. (Nor have I seen quoted \r or \n either...as far as I know.) Anyway, I think this request is worth a discussion and we should try to avoid getting distracted by the craziness of CTCP here.

[SadieCat]

Has anyone been working on this?

[SadieCat]

@grawity Do you think it would be would be possible for you to extract your tests to a separate repository (ircv3/test-vectors?) and convert them to a standard format (e.g. YAML, JSON, etc) that they can be consumed easily?

[DanielOaks]

I've started working on something like this over here, along with a little script to test them in the repo.

[Mikaela]

Does this issue include SASL REAUTH looks interesting, but I have no idea how to test it and I am too lazy to setup my own IRCd and services?

[kaniini]

No.