|
Cro: libraries and tools for building reactive services in Perl 6 | cro.services/ | Logs: irclog.perlgeek.de/cro/ Set by moderator on 17 January 2018. |
|||
|
01:02
SubxN8 joined
|
|||
| SubxN8 | Hi | 01:02 | |
|
01:05
SubxN8 joined
01:07
SubxN8 left
02:58
ilbot3 joined
|
|||
| moderator | Cro: libraries and tools for building reactive services in Perl 6 | cro.services/ | Logs: irclog.perlgeek.de/cro/ | ||
| Geth | cro: 0bacbc1358 | Altai-man++ | docs/http-auth-and-sessions.md Fix redirect |
10:38 | |
| cro-http/push-promise: 30 commits pushed by Altai-man++ review: github.com/croservices/cro-http/co...1a0e42224f |
11:31 | ||
| cro-http/push-promise: 6e7a7f5ec4 | (Jonathan Worthington)++ | 3 files Refactor handling of push promises in responses Get the response parser to attach them, which is simpler. This fixes the hang in the client getting push promises (which was caused by the client itself tapping the preserved Supply of them and thus eating them, so the end user never got to see them). Along with this, turn the test file exhibiting the hang into some basic push promise tests. |
13:28 | ||
|
14:31
lizmat joined
|
|||
| Geth | cro-http/push-promise: 8ad4cf8c36 | (Jonathan Worthington)++ | lib/Cro/HTTP2/Frame.pm6 More detailed HTTP/2 frame trace output |
14:37 | |
| cro-http/push-promise: 95ea6b5969 | (Jonathan Worthington)++ | lib/Cro/HTTP2/FrameSerializer.pm6 Correct property name to match usage Cro::HTTP::Client passed it with this name, not with -promise at the end. |
14:38 | ||
| cro-http/push-promise: 837a08c7b8 | (Jonathan Worthington)++ | lib/Cro/HTTP2/GeneralParser.pm6 Remove unrequired when/proceed |
14:42 | ||
| cro-http/push-promise: df17bdac62 | (Jonathan Worthington)++ | 2 files More properly handle HTTP/2 reset stream frames Including breaking the Promise associated with any HTTP/2 push promises that did not yet receive a response. |
14:51 | ||
| cro-http/push-promise: ac77e5d897 | (Jonathan Worthington)++ | it/akamai-push-promise.t Add basic integration test for push promises Which, happily, passes. |
15:29 | ||
| jnthn | Yay, push promises in the HTTP client actually work against a non-Cro HTTP/2 server :) | 15:30 | |
|
15:54
ilbot3 joined
|
|||
| moderator | Cro: libraries and tools for building reactive services in Perl 6 | cro.services/ | Logs: irclog.perlgeek.de/cro/ | ||
| timotimo | sweet! | 15:56 | |
| Geth | cro-http: 5a7b913b09 | Altai-man++ | 8 files Implement JWT-based Auth with tests; Also reformat META6.json to use 4 white space characters. |
17:37 | |
|
17:42
sena_kun joined
17:49
lizmat joined
|
|||
| Geth | cro: ef99ce300f | Altai-man++ | docs/http-auth-and-sessions.md Document JWT-based Auth classes |
18:02 | |
| jnthn | Hm, we ended up with JWT stuff in core. I initially thought a module. | 20:32 | |
| Though...they're probably common enough to include that battery. | |||
| sena_kun | Ah. | 20:55 | |
| A module. | |||
| :/ | |||
| Well, we can extract stuff easily. | 20:56 | ||
| I mean, it's just one file plus META6.json. | |||
| jnthn, ^ | |||
| jnthn | sena_kun: Let me think about it a little. I did question whether to include it. | 20:57 | |
| sena_kun | ok. | ||
| jnthn | My guideline is roughly "is this thing likely to be relevant for a long time" and "is this thing likely to be useful to a non-tiny number of users" | 20:59 | |
| sena_kun | btw, I've patched push promises. It indeed passes tests(e.g. switch variable works), but on disable the amount of settings frames increases. Investigating... | ||
| jnthn | Oh, and "do I want to dependency" :) | ||
| s/to/the/ | |||
| JWTs are basically JSON (will live forever, effectively) and a choice of crypto algorithm, the choice meaning that they're not tied to one algo that could become a less-than-good practice in the future | 21:00 | ||
| And they seem to be in widespread use already | |||
| And I guess we already have a dep on OpenSSL for, well, SSL support | 21:01 | ||
| So in that sense it's not really a new dependency | 21:02 | ||
| So, I'm inclined to leave it in core, and just provide that battery :) | 21:03 | ||
| sena_kun | cro doesn't present itself as a tiny project, does it? I mean, it's surely modular and stuff, but I don't see it as "We were able to save 4kb without that dependency". | ||
| jnthn | No | 21:04 | |
| I worked with express.js and, while I was in some ways impressed from a design perspective how very little they put in core...it was less fun when actually building something and having to go hunt the plugin for "basic" things | |||
| So not being that minimal in core was a design decision I made quite early on :) | 21:05 | ||
| At the same time, we don't want every possible thing we can think of in there :) | 21:06 | ||
| So it's just a case of considering each one. | |||
| And whether it's something that's going to be relevant to a decent number of our users (and potential future users), and whether it's likely to not be out of date/fashion in a year or two's time. | 21:07 | ||
| sena_kun | ugh. those suspicious settings frames are suspicious. :/ | 21:09 | |
| jnthn | By definition ;) | ||
| jnthn bbiab | |||
| sena_kun | exactly | ||
| oh, fixed it. | 21:12 | ||
| jnthn | yay :) | 22:22 | |
| Geth | cro-http/push-promise: 1b875b0629 | Altai-man++ | 2 files Remove Ping&Settings handlers from GeneralParser FrameParser sends it immadiately instead of passing further(i.e. to GeneralParser), so it's a dead code now, and the test was obsolete. |
22:25 | |
| cro-http/push-promise: 8ac2900ecb | Altai-man++ | 5 files Refactor server to avoid pushing when disabled Initially done in two phase handshake(first preface with Bool, then actuall Settings) does not really make much sense, so we can just emitting a Settings frame from the beginning. In response to that, we negotiate server settings based on client's one and send server's settings, then acknowledged settings frame(according to RFC). Fix a tests that were broken by this change and make test's check better, covering the case when FrameSerializer emits more frames than we are awaiting. |
|||
| sena_kun | new commits. :) | ||
| sena_kun | hmm. I tried to explain situation as good as I could. Maybe I went overboard a bit. :P | ||
| jnthn | No, those are good commit messages. :) | 22:26 | |
| sena_kun | unfortunately, did not manage to test things with browser today. *yesterday already. | ||
| jnthn | That's fine, can happen tomorrow | 22:27 | |
| sena_kun | *today | ||
| :) | |||
| Ah, it's minus one hour over there. | |||
| :D | |||
| jnthn | Yeah, it's still today here ;) | ||
| Time. How does it even work. :-) | |||
| sena_kun | >google sheets | 22:28 | |
| >-10 hours | |||
| Ugh. | |||
| Splitting it. | |||
| jnthn | haha...I've had that too | ||
| I think it's google's gentle way of telling people who work at midnight, maybe you should sleep :-) | |||
| Hm, seems I put "Generalize body parsing and provide it for web sockets messages and ZeroMQ messages" into the 0.8 goals, so maybe will take a look at that at least for websockets tomorrow | 22:29 | ||
| (Next release will be 0.7.3 for sure) | 22:30 | ||
| I should dive into Cro::ZeroMQ stuff soonish too | 22:31 | ||
| sena_kun | ewww. | ||
| I'm afraid to touch it because of 4.1/4.2 situation. | 22:32 | ||
| jnthn | Oh, right, that means there's every chance I'll be like "works for me!" and half the rest of the world will be like "lol, nope" : | 22:34 | |
| :/ | |||
| Still, with our HTTP[S|2.0] in increasingly good shape, I'd like to take a look at getting the ZeroMQ support in better shape | 22:35 | ||
| sena_kun | it's worse that we don't really know where the bug is. But we can try to find that out for sure. | ||
| jnthn | Yeah, I'll see what I can do | ||
| jnthn should get a "Debugger Of Last Resort" t-shirt or something | |||
| sena_kun | with superman's "S", perhaps. :) | 22:36 | |
| jnthn | :P | 22:37 | |
| sena_kun | ok, it's time to sleep already. o/ | 22:55 | |
| jnthn | Rest well, thanks for all the work o/ | 22:56 | |
|
22:57
lizmat joined
|
|||