#twisted-dev

/

      • stryx` joined the channel
      • wirehead_ has quit
      • wirehead_ joined the channel
      • adiroiban joined the channel
      • adiroiban has quit
      • oberstet2 joined the channel
      • kenaan
        Tickets pending review: https://tm.tl/#8009 (Unit03), #8966 (the0id), #9006, #9032, #9040, #9042, #9047, #9066, #9064 (emattiza), #9072, #9083 (rodrigc), #9084, #9086, #5310, #4819
      • runciter
        glyph: if you've got a spare moment to review: https://github.com/twisted/treq/pull/178
      • should twm take a look too?
      • once that's good, we can get https://github.com/twisted/treq/pull/177 in
      • glyph: ty :)
      • glyph
        heyo
      • runciter
        buh
      • uh
      • i renamed that so sphinx would actually show me the rendered docstrings
      • forgot to unrename it
      • glyph: and just like that the problem disappeared
      • glyph
        runciter: A+
      • runciter: well uh
      • runciter: see comment.
      • runciter
        how could it ever work
      • glyph: presumably this worked because twisted depends on attrs, but presumably this doesn'
      • t work because attr gets imported as attr?
      • who cares - fixed!
      • glyph
        runciter: I think it's probably some accident of implementation in pip
      • runciter
        i'll buy that
      • glyph: so the HTTPS issue
      • is there an issue?
      • it's weird that TCP4ClientEndpoint gets used, but it's actuall irrelevant to the operation of HTTP protocols in RequestTraversalAgent
      • glyph
        runciter: .isSecure().
      • runciter: This is actually going to be relevant when I need to test the rest of form-hacks, because it has different behavior based on that flag :)
      • runciter
        it's only used to a) retrieve the port back from the parsed URI and b) to set up a deferred that fires when the client connection's established
      • glyph: ah yes it does
      • as in "works" or "doesn't work" as i remember form-hacks
      • glyph: buh, so it calls _channel_ isSecure, which asks the transport if it provides ISSLTransport
      • ok, got it
      • glyph
        runciter: you can possibly fake this a lot
      • runciter: or you could just set up trust roots and do a for-real TLS connection in memory
      • totally your call :)
      • runciter
        glyph: http://i.imgur.com/jp0Bbnv.png is why i do not want to in-memory TLS
      • this is a moto test
      • glyph
        is that jitviewer?
      • runciter
        just vmprof!
      • with cpython, actually, for boring reasons
      • vmprof is *good*
      • uh
      • what is travis doing
      • tomprince
        runciter: What https issue are you discussing?
      • There is https://github.com/markrwilliams/treq/blob/275f... so that isSecure does the correct thing.
      • At least, that is the point of it, I haven't checked if it works.
      • runciter
        glyph: is that broken?
      • tomprince: that certainly looks like it's trying to satisfy HTTPChannel.isSecure
      • tomprince
        Note that in any case, the transport is always an iosim.FakeTransport (both before and after)
      • The only difference is what attribute on the MemoryReactor the factory is stored in.
      • kenaan
        glyph reviewed <https://tm.tl/#9006>; - successResultOf and failureResultOf should accept coroutines (assigned to wsanchez)
      • adiroiban joined the channel
      • runciter
        tomprince: you the only difference between the code before and after this PR?
      • you mean*
      • tomprince
        Well between http and https in the old code and/or between what the old and new code do for https
      • So, yes.
      • glyph
        Oh.
      • Well nevermind then great
      • Land pls
      • glyph joined the channel
      • cyli joined the channel
      • runciter
        ok i'm going to close and reopen the PR in an attempt to wake up travis
      • i doubt that did much
      • oberstet2 joined the channel
      • lukasa
        Hey twistedeers, is updating a dependency considered an enhancement, or a misc?
      • kenaan
        new web enhancement https://tm.tl/#9096 by Lukasa: Move to hyper-h2 3.0
      • KenaanX joined the channel
      • NOTICE: [13twisted] 15Lukasa created 069096-lukasa-update-h2-dependency (+2 new commits): 02https://git.io/vSL6A
      • NOTICE: 13twisted/069096-lukasa-update-h2-dependency 143366f20 15Cory Benfield: Move to h2 3.0
      • NOTICE: 13twisted/069096-lukasa-update-h2-dependency 14eb1a72d 15Cory Benfield: Add topfile for #9096
      • KenaanX has left the channel
      • KenaanX joined the channel
      • NOTICE: [13twisted] 15Lukasa opened pull request #755: Update h2 dependency to 3.0. (06trunk...069096-lukasa-update-h2-dependency) 02https://git.io/vSLif
      • KenaanX has left the channel
      • Lukasa submitted <https://tm.tl/#9096>; - Move to hyper-h2 3.0 (unassigned) for review
      • lukasa
        hawkowl: I got you a giiiiift ^^^
      • tomprince
        lukasa: I'd definitely not say misc, since it might cause issues if people also have a direct dependency on a thing they can't upgrade.
      • kenaan
        Tickets pending review: https://tm.tl/#8009 (Unit03), #8966 (the0id), #9032, #9040, #9042, #9047, #9066, #9064 (emattiza), #9072, #9083 (rodrigc), #9084, #9086, #5310, #4819, #9096
      • clokep joined the channel
      • clokep joined the channel
      • adiroiban has quit
      • adiroiban joined the channel
      • clokep joined the channel
      • Tickets pending review: https://tm.tl/#8009 (Unit03), #8966 (the0id), #9032, #9040, #9042, #9047, #9066, #9064 (emattiza), #9072, #9083 (rodrigc), #9084, #9086, #5310, #4819, #9096
      • stryx` joined the channel
      • wsanchez
      • Twisted CI is already pretty bloated. Does Twisted really need to be testing downstream projects in its own CI?
      • glyph
        wsanchez: well, it's only a _little_ downstream
      • wsanchez
        Sure. Why not Klein?
      • And txaws
      • And rxmongo
      • And tubes
      • glyph
        wsanchez: OK so, eventually, we want that. If we break a downstream, we want to know. But, of course, having them all in our build matrix would be ridiculous.
      • wsanchez: I think having 2-3 of these as a stopgap is what we're aiming for as a stopgap, and by the time we get around to adding more, we should be figuring out something more elaborate so we get a notification somewhere that shows up and gets paid attention to, but doesn't necessarily have a gigantic blocking thing on CI
      • but like
      • "we don't break treq on one platform" (which has a very fast, very small test suite) seems like a reasonable bar for the time being.
      • wsanchez
        I agree that we want to know, and certainly before a release. But including them in the main CI runs seems nuts to me.
      • glyph
        wsanchez: If they all ran in parallel (== travis gave us infinite money) would that still seem crazy?
      • runciter
        how do i do a treq release?
      • looks like it's using incremental, so i know how to bump the version
      • but it doesn't look like treq uses tags (or tracks releases on branches)
      • i also don't have the pypi credentials
      • i'm perfectly happy not to do any of this if somebody _else_ wants to do a release
      • but given that https://github.com/twisted/treq/pull/177 is now merged, and treq works with twisted 17.1.0 again, it seems like it's time to release
      • glyph
        runciter: let me pypi you. runciter?
      • runciter
        glyph: my pypi username is markrwilliams
      • glyph
        runciter: you are now an authorized uploader on treq, klein, and, what the heck why not, automat
      • runciter
        glyph: ho boy
      • i'm rewriting automat in cython
      • microbenchmarks, ya know
      • glyph: any opinions on releasing treq?
      • as in: would you like a tag?
      • that sort of thing
      • i'll follow moshez's fantastic lead and write a RELEASING doc
      • glyph
        runciter: Yes. Please add a tag, and also that thing about documentation.
      • runciter: The existing tags look super inconsistent, unfortunately.
      • runciter: Thanks!
      • (Back to work for me.)
      • runciter
        glyph: want a branch too? twisted style?
      • that's probably not necessary because we'll just roll forward, yeah?
      • glyph
        runciter: you're the release manager; via con dios
      • runciter
        glyph: 👍
      • what could go wrong
      • glyph: have a good one
      • glyph
        runciter: you too :)
      • wsanchez
        glyph: If Travis was super fast I suppose it doesn't matter but that's not anywhere near my experience.
      • Though if I worked at Travis I'd be throttling the Twisted org for heavy use of the free service.
      • By which I mean I'd make the scheduler prioritize projects/orgs that are less expensive.
      • stryx` joined the channel
      • oberstet joined the channel
      • stryx` joined the channel
      • kenaan
        Tickets pending review: https://tm.tl/#8009 (Unit03), #8966 (the0id), #9032, #9040, #9042, #9047, #9066, #9064 (emattiza), #9072, #9083 (rodrigc), #9084, #9086, #5310, #4819, #9096
      • stryx` joined the channel
      • average has quit
      • average joined the channel
      • adiroiban has quit
      • adiroiban1 joined the channel
      • adiroiban1 has quit
      • itamar joined the channel