#xi

/

      • sleffy has quit
      • sleffy joined the channel
      • mysangle joined the channel
      • mysangle has quit
      • mysangle joined the channel
      • mysangle has quit
      • mysangle joined the channel
      • bemeurer joined the channel
      • bemeurer has quit
      • mysangle has quit
      • mysangle joined the channel
      • mysangle has quit
      • mysangle joined the channel
      • sleffy has quit
      • thvdburgt joined the channel
      • mysangle has quit
      • thvdburgt has quit
      • shymega has quit
      • shymega joined the channel
      • shymega has quit
      • shymega joined the channel
      • Yaniel has quit
      • raph
        xray, hmm
      • Lnavry is now known as
      • Yaniel
        finally got my nick back \o/
      • should probably register it
      • cmyr
        :)
      • xray? what'd i miss
      • raph
      • cmyr
        interesting
      • I knew they were thinking about rust a lil, didn't know it was this major
      • raph
        i don't have a strong sense of what's actually going on there
      • i wonder if any of the people are local to the bay area, at this point having a chat would be helpful
      • cmyr
        totally
      • theoretically there's lots of room for shared work
      • raph
        this is all nathan sobo, who's in boulder
      • electron is undoubtedly the biggest problem with the project
      • it precludes solving startup time and ram usage, which i perceive as the biggest opportunities wrt atom
      • in any case, to the extent possible i'd like to share tech, it's occurred to me a long time ago that we could share syntax highlighting with other editors
      • (to me, that idea makes even more sense as we start building better highlighters)
      • cmyr
        they have a new highlighting project, but it's cpp
      • raph
        if you're gonna inject rust, that's a really good place for it
      • anyways
      • i'm doing ok, but merely ok
      • feel like i lost a lot of momentum over the past few days
      • family crisis has simmered down to low, thankfully
      • cmyr
        no worries, focus on work, get your talk together
      • raph
        yeah, will absolutely do that
      • just all the extra goodies feel like they're slipping
      • cmyr
        i'm mostly working on a few other little things right now
      • like which?
      • raph
        finishing up the mac perf work, like fake incremental present
      • haven't really started
      • cmyr
        well fwiw performance right now is absolutely acceptable
      • I'm using the editor regularly
      • raph
        also when things get tight, external comms feel like they take the hit
      • cmyr
        also the little hacky xi.sh script i posted in xi-mac is a nice treat
      • from vim I can just do !xi %
      • raph
        i'm reminded a bit of philip guo's talking/doing ratio (http://pgbovine.net/PG-Vlog-16-talking-to-doing...)
      • heh nice
      • cmyr
        yea no worries about comms, if I'm ever blocked I'll actively annoy you
      • raph
        thanks
      • in any case, will be talking next week, so that hopefully will get the ratio back in whack
      • cmyr
        yea nice. I'm going to figure out this morning actually if I'll make it down to nyc
      • raph
        i think i'm going put a bigger emphasis on perf (in the talk), less about "advanced" stuff
      • one idea i'm leaning toward is showing some screenshots from the chrome tracing
      • cmyr
        yea if it's 30 minutes focusing on something specific seems like a good approach
      • raph
        when i gave an earlier draft of the talk it took about 50 minutes, but i had q&a interleaved
      • this time the format is just me speaking, with (non-recorded) q&a after
      • cmyr
        it'll fly by
      • raph
        dan luu solved this by talking about 2x speed :)
      • (this is a tokio impl of xi rpc)
      • cmyr
        I looked at it a while ago, I don't really know enough about tokio to really have a well-formed opinion?
      • it seemed well put together though
      • raph
        coolness, i'm inclined to say yes here
      • cmyr
        I was talking to littledude about it a while ago, definitely when we move to tokio for core I want to have a shared crate for the API
      • and xrl seems like a good base for that
      • I _think_ it only really has the client side, currently?
      • raph
        i haven't looked in detail yet
      • i think if one side is fully built out doing the other is not that hard
      • cmyr
        yep
      • raph
        windows apps have a very complicated story under the hood for threading / async
      • you've got the main message pump, which you need for things like input and window resize
      • then you have things like "wait objects" for presentation, and those kinda want to be on another thread
      • i never really worked out in xi-win how all these pieces should fit together, especially as i went back and forth on gamelike vs layered
      • cmyr
        I've been more or less putting my head in the sand re: xi-win, although this time of year having a windows install becomes tempting so I can play games :}
      • raph
        it makes sense for it to be on hold right now
      • though in the slightly longer term i'm very excited about its prospects
      • by coding it in rust on top of winapi it has the potential to be incredibly lightweight and performant
      • cmyr
        yea staying in rust is compelling
      • kk i'm going to grab a bite, back in a bit
      • raph
        sure thing, enjoy
      • the flickery behavior on window resize seems to be identical in 10.13 as it was on 10.12, maybe i should file a radar
      • i followed up a little on xi-win#11
      • cmyr
        yea, makes me think there's a bunch of stuff in my head about the runloop refactor/asyncio that we've talked about but don't have issues for
      • raph
        it's certainly going to overlap with a lot of things in flight
      • like the timeout to aggregate plugin-generated changes if they come in quickly enough
      • cmyr
        yea it touches a lot of stuff
      • raph
        that said, i definitely see the value in starting with the front-end before tackling core
      • the windows stuff i brought up there is tricky
      • when i was researching the issues i saw a lot of evidence that people struggled with it
      • cmyr
        ah just reread #11
      • hadn't read the initial question carefully
      • raph
        stupid rust question - if i set `default = "notify"` for example, is there an invocation to disable all features?
      • passing `--features ""` to cargo doesn't seem to do that
      • (and that should be `default = ["notify"]` sorry)
      • cmyr
        is there a --no-defaults flag?
      • I thought there was something like that
      • yea, --no-default-features
      • raph
        got it
      • it seems confusing though; i'm more inclined to have everything off by default but opt-in
      • cmyr
        I can see the argument on the other side
      • where there's a configuration you expect to be used in 99% of cases, but some people may need to disable some component
      • not sure I have strong feelings either way..
      • raph
        actually i think i'll go with on-by-default
      • sleffy joined the channel
      • sleffy has quit
      • log 0.3.9 depends on log 0.4?
      • file under "strange but true" i think
      • Yaniel
        isn't that called the 'semver trick'
      • to implement breaking changes without breaking dependent crates
      • raph
        from a quick scan, it seems to be to support no-std specifically
      • i'm auditing the crates xi depends on (including transitively) in part because we nominally have a policy of only one version per crate
      • but stuff like this basically makes that policy impossible to enforce
      • Yaniel
        iirc this is to be able to change type definitions
      • libc::c_void was named one example
      • raph
        (should probably be having this discussion on fuchsia-rust rather than here :) )
      • cmyr has quit
      • cmyr joined the channel
      • vlovich joined the channel