[7:12]<Dantman> I only have one regret now... I'm writing a nice fancy system with uploads, filter based files, thumbnails, background transcoding of video into multiple formats, and more fun... But I probably won't get to show off the code... [9:06]* Dantman can't decide if he should serve uploaded text files with text/plain or serve them with an invalid text/text to avoid IE's extension guessing. [18:25]<Wes-> I can't believe I've been working on my "simple" idea for unicode buffers on binary/f for three hours [18:25]<Wes-> holy crap, time flies [18:26]<ashb> heh [18:26]<ashb> a text filter on top of them? [18:42]<Wes-> ashb: no - I think we can have a normal API and a fast API, where normal API supports iconv and fast API is unicode-only [18:43]<Wes-> fast API can work buffer-to-buffer, supporting partial characters [18:43]<Wes-> normal API intended work on line data [18:43]<ashb> i see [18:43]<Wes-> if you want partial character (character stream) support, you need the full encoding spec [18:44]<Wes-> sticking with only unicode has real perf advantages [21:09]<Dantman> T_T My <script> progress has one fault... For some reason when I upload a large file and try to abort it sending a document with a script tag it doesn't get executed. Works for small files though. [21:28]<devinus> ryah: is there a node channel? [21:28]<devinus> anymore? [21:29]<Wes-> devinus: #nodejs IIRC [21:30]<devinus> hrm [21:30]<devinus> used to be a lot more people in there [21:30]<Wes-> hmm, I see what you mean [21:31]<devinus> there's also no topic set anymore [21:31]<devinus> just weird in general [21:32]<ryah> devinus: #node.js [21:32]<devinus> aha! [21:32]<devinus> thanks, ryah [21:33]<devinus> -, [21:33]<Wes-> I was just about to say that [21:33]<Wes-> (had to RTFM :) ) [21:33]<MisterN> ryah: you could make #nodejs redirect to #node.js [21:33]<devinus> JohnResig is in the fake channels #ndoe and #nodejs :-P [21:33]<devinus> it's sort of....creepy [21:34]<ryah> MisterN: oh, how do i do that? [21:35]<MisterN> ryah: oh i don't know how precisely to do it on this network, but i know it's possible. you need to read the docs :) [21:35]<Wes-> OMG, John is typo-squatting! [21:36]<MisterN> ryah: on euirc, i can ban everybody and set up a ban-redirect [21:37]<MisterN> you also need to make sure the channel won't be dropped so you need somebody to idle in there (maybe a bot, or somebody with a bouncer) [21:37]<devinus> what a hassle [21:41]<Dantman> ryah, Register with chanserv, set a mode +if #otherchan and put it in the MLOCK. (+f forwards users to the channel if they can't join it, and +i makes it so that everyone gets forwarded since they aren't invited) [21:41]<Dantman> You'll need to be +o in both channels. [21:42]<ryah> oh okay [23:53]<Wes-_> ryah: ping [23:56]<ryah> Wes-_: pong [23:57]<kriskowal> pang [23:57]<Wes-_> ryah: have you ever considered stealing the underlying buffer from a UTF8-encoded-v8 string and sliding it into a Binary/F-style buffer? [23:57]<Wes-_> ryah: I was thinking - it would be easier to allow that if we have an immutable buffer type, but on reflection, CoW is probably easier/saner [23:58]<Wes-_> ryah: I am considering similar for spidermonkey UTF16 strings [23:58]<ryah> Wes-_: utf8-encoding-v8 string? [23:58]<Wes-_> ryah: Although with spidermonkey you have to be careful because strings aren't *actually* immutable [23:58]<ryah> Wes-_: i think all utf8 strings are transformed to 16bit internally [23:58]<Wes-_> ryah: utf8-encoded, sorry [23:59]<ryah> Wes-_: so i don't think there is ever an underlying utf8 buffer [23:59]<Wes-_> ryah: I have only read the source once -- quickly -- but am definitely under the impression v8 will store strings as utf8 given the opportunity [23:59]<Wes-_> ryah: presumably using charAt() etc cause a utf8->16 marshall [23:59]<ryah> it definitely does that for ascii [23:59]<Wes-_> ryah: was thinking if you do lots of utf8 content output in node, this could be a good savings for you