[4:11]<Dantman> ... parsing multipart/form-data file uploads... /fun/... [7:47]<ondras> so [7:47]<ondras> some SM embedder here? [8:15]<Dantman> heh... http://pastie.org/841951 [8:16]<ondras> I recall swearing a lot when I implemented multipart parsing in v8cgi [8:28]<DanielFriesen> ^_^ This is going to be fun... I'm going to use multipart/form-data in a way that only one field is accepted... instead of putting an upload id into another part of the multipart message the upload id is going to be located in the field name which will end up inside of the Content-Disposition for the file... [8:29]<DanielFriesen> No wait... screw that, I forgot, I wasn't even going to make the client generate an upload ID... I'll just send one in the kuploadfinished event. [8:52]<DanielFriesen> Reading over the persistent connections part of the HTTP spec makes me want to replace JSGI with a HTTP interface that makes the difference between a response buffered in the app and a response which is streamed and terminated by close explicit so that custom http servers can allow persistent connections. [8:53]<DanielFriesen> Hmm, then again Content-Length might make that not necessary [9:13]<DanielFriesen> hmmm... useful stream and buffer combination api.. [9:23]<DanielFriesen> var ascii = "US-ASCII"; var CRLF = "\r\n".toBlob(ascii); while( sb.indexOf(CRLF)) >= 0 && !sb.eof ) sb.extend(512); var hline = sb.range(0, sb.indexOf(CRLF)).toString(ascii); sb.shift( [9:23]<DanielFriesen> Ack, damn laptop [9:37]<DanielFriesen> *sigh* Java's java.utils.Arrays is almost useless [9:40]<DanielFriesen> T_T Thank you beautiful Mozilla JS array generics... [11:09]<DanielFriesen> *twitch* Aaaagh... did I just spend hours trying to parse a mime type and it's parameters and then realize that my .scan combined with a regex makes that a one liner? [11:17]<DanielFriesen> Wow, Chrome is fairly abusive of /favicon.ico