2004/09/14
IRC [01:34] *** zoro2 joined the chat.
IRC [01:45] *** zoro2 parted the chat.
IRC [03:38] *** frankie joined the chat.
IRC [07:19] *** erph parted the chat.
IRC [07:20] *** erph joined the chat.
IRC [11:19] *** aolserver-scribe parted the chat.
IRC [11:35] *** frankie parted the chat.
IRC [12:04] <jhavard> busy morning
IRC [12:11] *** archiware joined the chat.
IRC [12:12] <archiware> howdy!
IRC [12:18] <jhavard> morning
IRC [12:19] <jhavard> err, good whatevertime
IRC [12:19] <archiware> good evening here
IRC [12:20] <jhavard> well, lunch time here.
IRC [12:20] * jhavard &
IRC [12:20] <archiware> bon apettite
IRC [12:21] <Dossy> okay, after talking to a co-worker, i think i'm going to work on my own tcl xml extension package :P
IRC [12:21] <Dossy> but, not today - need to leave the office soon
IRC [12:21] <archiware> yatclxml
IRC [12:22] <Dossy> heh yup
IRC [12:22] <Dossy> except mine will act a lot more tcl-ish
IRC [12:22] <archiware> tdom wont do the work?
IRC [12:22] <Dossy> no, i'm not too keen on tdom or nsxml --
IRC [12:22] <Dossy> and IIRC, TclXML is a pure-tcl impl.
IRC [12:22] <archiware> any particular reason?
IRC [12:23] <Dossy> I want an xml package that goes back to the roots of Tcl's "everything is a string"
IRC [12:23] <archiware> ?
IRC [12:23] <Dossy> the basis of my extension will be to define a new Tcl_Obj type for XML
IRC [12:23] <Dossy> Do things like:
IRC [12:23] <Dossy> set xml "<foo><bar>baz</bar></foo>"
IRC [12:24] <Dossy> xml $xml xpath /foo/bar
IRC [12:24] <Dossy> and it'll return a Tcl_Obj of type xmlObjType which refers to the text node "baz"
IRC [12:24] <Dossy> then, if I use it in a string context (i.e., "puts $obj") it'll shimmer it back to its string rep ("baz")
IRC [12:25] <Dossy> But, I could say:
IRC [12:25] <Dossy> set node [xml $xml xpath /foo]; puts [xml $node children] and it'll output 1.
IRC [12:25] <Dossy> $xml will get parsed by the first [xml] call because it's of Tcl_Obj type stringObjType
IRC [12:26] <Dossy> but the second [xml $node ...] call won't do any XML parsing because $node is already of type xmlObjType
IRC [12:26] <Dossy> the idea is that Tcl will handle shimmering from string, to list, to XML internal rep (via xml parser) as necessary. Just like Tcl does for its other data types.
IRC [12:27] <Dossy> This also means that when Tcl_Obj's go out of scope, they get freed/destroyed. Ref counting will handle copy-on-write or value-replacement as appropriate, etc.
IRC [12:27] <archiware> i recall Rolf wanted to do similar but putting the dom tree underneat the Tcl _Obj
IRC [12:28] <Dossy> yeah
IRC [12:28] <archiware> i think this should be very fast
IRC [12:28] <Dossy> if tDOM goes that direction, great. otherwise, I'll build mine and see if people take to it.
IRC [12:29] <archiware> well, can't say much now, but there are some problems with shared dom trees
IRC [12:29] <Dossy> but I know there's PLENTY of people already using tDOM, and this kind of fundamental design change can't be thrust upon them
IRC [12:29] <archiware> nono...
IRC [12:29] <archiware> I think your idea is ok
IRC [12:29] <archiware> i'm just grumbling now..
IRC [12:30] * Dossy nods
IRC [12:30] <archiware> i know we wanted to do such thing, but then people wanted shared dom trees
IRC [12:30] <archiware> which aint easy to build using tcl_objs in mt-process
IRC [12:30] <Dossy> right
IRC [12:30] <Dossy> need a real mechanism for sharing tcl_obj's across threads.
IRC [12:31] <archiware> correct.
IRC [12:31] <archiware> this is a huge drawback of tcl
IRC [12:31] <Dossy> problem is shimmering which would lose the "thread-shared" flag
IRC [12:31] <Dossy> because you can't really represent it in the string rep. :(
IRC [12:32] <Dossy> anyhow, open bugs at sourceforge is down around 35 now :)
IRC [12:32] <Dossy> i stand a good cahnce of closing them ALL by the end of 2004 ...
IRC [12:32] <Dossy> or at least, getting it down to no more than 20 open bugs
IRC [12:32] <archiware> i've seen... you've been busy lately :)
IRC [12:33] <archiware> off topic... do you guys know or use skype for internet telephony?
IRC [12:40] <Dossy> nope, not me
IRC [12:40] <archiware> so nobody...
IRC [12:40] <archiware> i'm just amazed how this thing works!
IRC [12:40] <Dossy> cool :)
IRC [12:41] <Dossy> as long as you have a fast, low-latency connection ... :)
IRC [12:41] <Dossy> i'm going offline for a bit - ttyl
IRC [12:41] <archiware> seeya
IRC [12:44] <archiware> dossy: you should really checout www.skype.com
IRC [12:45] <archiware> i've been talking to my girlfiend from Japan (to Germay) four *hours* w/o any cost
IRC [12:45] <archiware> for the comparison: a 5 minute telephone call costed about 10 bucks
IRC [12:46] <archiware> i love skype. I just love it.
IRC [13:13] *** tekbasse joined the chat.
IRC [13:14] *** tekbasse parted the chat.
IRC [13:17] *** frankie joined the chat.
IRC [13:39] <jhavard> archiware: that's why voip must be regulated. It doesn't let the monopolies make money
IRC [13:41] <archiware> i'm not that knowledgeable on that matter but isn't skype voip?
IRC [14:58] <jhavard> yep.
IRC [14:58] <jhavard> Which is why the government must regulate... so the monopolies can control it.
IRC [15:00] <jhavard> otherwise, public safety is at risk.
IRC [15:01] <jhavard> (no, I don't believe a word of it either)
IRC [15:28] <jhavard> Is the taglib in head?
IRC [16:38] *** zoro2 joined the chat.
IRC [17:34] *** frodoroot joined the chat.
IRC [17:35] <frodoroot> hello?
IRC [17:39] *** frodoroot parted the chat.
IRC [17:55] *** frankie parted the chat.
IRC [18:42] <Dossy> jhavard: no, taglib is not in HEAD
IRC [18:42] <Dossy> I'm hesitant to commit it until people give it a review.
IRC [18:43] <Dossy> I'm about to file a RFE to implement the "<tag/>" notation support change into HEAD and 4.0.9, though.
IRC [18:50] *** zoro2 parted the chat.
IRC [20:02] <Dossy> So, who's still here?
IRC [20:41] *** brunom joined the chat.
IRC [20:41] <Dossy> 'lo, Bruno.
IRC [20:53] <brunom> Hello Dossy
IRC [23:11] *** Alexol joined the chat.
IRC [23:11] *** Alexol parted the chat.
IRC [23:13] *** martinh joined the chat.