#exult@irc.freenode.net logs for 26 Nov 2002 (GMT)

Archive Today Yesterday Tomorrow
Exult homepage


[00:33:53] --> Kirben has joined #exult
[00:33:53] --- ChanServ gives channel operator status to Kirben
[00:41:53] <wjp> hi Kirben
[00:42:00] <Fingolfin> hi
[00:42:09] <Kirben> Hi
[00:42:14] * wjp hmms... way too late (again)
[00:42:18] <wjp> I guess I should be going :-)
[00:42:21] <Fingolfin> wjp: "lichaamsuitbreiding" is a "field extension", right?
[00:42:26] <wjp> Fingolfin: yes :-)
[00:42:26] <Fingolfin> wjp: yeah, way too later
[00:42:34] <Fingolfin> wjp: unfortunatley, I have this deadline for wednesday <g>
[00:42:38] <Fingolfin> wjp: thx
[00:42:42] <wjp> I have to give a talk about them tomorrow :-)
[00:42:52] <wjp> (field extensions, that is)
[00:43:21] <Fingolfin> ah hehe
[00:43:21] <wjp> the proof of Kronecker-Weber, specifically
[00:43:21] <Fingolfin> ok
[00:43:37] <Fingolfin> in this case it's field extensions used for symbolic integration
[00:43:43] <Fingolfin> but it's not actually related to my deadline =)
[00:43:52] <Fingolfin> it's more for the oral examination next week :-)
[00:44:17] <wjp> field extensions and symbolic integration? hm, not sure I see what you mean
[00:44:41] <Fingolfin> well
[00:44:45] <Fingolfin> the risch algorithm
[00:44:56] <wjp> *blank* :-)
[00:44:57] <Fingolfin> you start by integrating polynomials
[00:45:04] <wjp> ok, I can do that :-)
[00:45:15] <Fingolfin> and then you extend your field by a new element, let's call it "ln(x)"
[00:45:29] <wjp> suggestive naming, but ok :-)
[00:45:29] <Fingolfin> we are actually on a differential field
[00:46:14] <Fingolfin> yeah, and that new element ln(x) (or "alpha" if you prefer), has the special property that D(alpha) = D(X)/x
[00:46:25] <Fingolfin> where D is the differential operator
[00:46:27] <-- Fingolfin has left #exult ()
[00:46:34] --> Fingolfin has joined #exult
[00:46:34] --- ChanServ gives channel operator status to Fingolfin
[00:46:37] <Fingolfin> oops, wrong key =)
[00:46:41] <Fingolfin> anyways
[00:46:41] <wjp> :-)
[00:47:04] <Fingolfin> the idea is, you don't treat ln or exp like function, but rather as abstract elements of your field with special properties
[00:47:15] <Fingolfin> as transcendental extension to be precises
[00:47:20] <Fingolfin> (so that they are not root of any polynomial)
[00:47:27] <Fingolfin> you can also add ln(1+x^2)...
[00:47:46] <Fingolfin> so the idea is, if you want to integrate something, you keep extending your field until you can integrate everything in your formula
[00:48:25] <Fingolfin> it's quite abstract, but using the risch algorithm, you can actually proof for a wide class of functions whether they are integratable or not, and what the integral is
[00:48:46] <Fingolfin> esp. the part where you can proof that there is *no* closed expression for the integral is nice...
[00:48:50] <wjp> I see
[00:49:12] <Fingolfin> I think the Risch algorithm (or improved version of it) are employed by various CAS, like Mathematica or Maple
[00:49:46] <Fingolfin> we only did the "basic" version of it, of course :-) you can extend it to trigonometric function, too, etc., but it's mostly techincal stuff, so not so much fun, after a while
[00:49:57] <Fingolfin> really something that you'd want to implement in a computer, not do manually :-)
[00:50:04] <wjp> I'm not sure I want to implement it :-)
[00:50:13] <wjp> you'd want someone else to implement it :-)
[00:50:14] <Fingolfin> hehe yeah
[00:50:28] <Fingolfin> but the paper I am working on is actually on automatic geometry theorem proving :-)
[00:50:44] <wjp> ah, cool
[00:50:50] <Fingolfin> actually, I am supposed to do it with another guy...
[00:50:56] <wjp> but he's MIA?
[00:50:59] <Fingolfin> so far, I wrote 90% of it... hrm...
[00:51:03] <Fingolfin> so to say =)
[00:51:16] <Fingolfin> haven't heard of him since the 12th (last mail to him dated back then)....
[00:51:22] <wjp> nice :/
[00:51:27] <Fingolfin> and our deadline is CLOSE
[00:51:45] <Fingolfin> I really hope he did do the parts he wanted to work on....
[00:52:27] <wjp> ugh, yeah
[00:52:30] <Fingolfin> it's not as if my life depends on this mark, but heck, this is still annoying. don't think I'll want to work with him on a paper again in the near future
[00:52:41] <wjp> sounds like a good idea, yes :-)
[00:53:16] <Fingolfin> though I must admit I have been slacking in the last two weeks, too... had to work on another paper, and do some work, and also was a bit lazy (relaxing and reading and coding on ScummVM etc.)
[00:53:26] <wjp> anyway, I should really be going now.. have to get up in 6 hours
[00:53:37] <Fingolfin> yeah :-)
[00:53:40] <Fingolfin> sorry for holding you back
[00:53:45] <Fingolfin> sleep well, and cya!
[00:53:52] <wjp> np, a couple of minutes less sleep won't hurt too much :-)
[00:53:59] <wjp> g'night and good luck on your paper :-)
[00:54:04] <-- wjp has left IRC ("Zzzz...")
[01:04:37] <-- Fingolfin has left IRC ("42")
[03:23:29] --> Kirben2 has joined #exult
[03:23:54] <-- Kirben has left IRC (Read error: 104 (Connection reset by peer))
[05:22:29] --- Darke|zzZ is now known as Darke
[08:54:28] --> Dominus has joined #exult
[08:54:33] --- ChanServ gives channel operator status to Dominus
[08:54:47] <Darke> Hi!
[08:55:03] <Dominus> hey ho
[10:43:12] <-- Dominus has left #exult ()
[10:48:28] --> Colourless has joined #Exult
[10:48:28] --- ChanServ gives channel operator status to Colourless
[10:48:52] <Colourless> hi
[10:50:50] <Darke> Hi.
[11:01:32] --> slacked has joined #exult
[11:16:38] <-- Darke has left IRC ("BRB!")
[11:24:37] --> Darke has joined #exult
[11:24:37] --- ChanServ gives channel operator status to Darke
[12:30:30] <-- Kirben2 has left IRC ("System Meltdown")
[12:41:28] --> Kirben has joined #exult
[12:41:28] --- ChanServ gives channel operator status to Kirben
[12:53:53] <-- Darke has left IRC (saberhagen.freenode.net irc.freenode.net)
[12:53:53] <-- slacked has left IRC (saberhagen.freenode.net irc.freenode.net)
[12:53:53] <-- Kirben has left IRC (saberhagen.freenode.net irc.freenode.net)
[12:53:53] <-- Colourless has left IRC (saberhagen.freenode.net irc.freenode.net)
[12:53:53] <-- cnj has left IRC (saberhagen.freenode.net irc.freenode.net)
[12:54:48] --> Kirben has joined #exult
[12:54:48] --> Darke has joined #exult
[12:54:48] --> slacked has joined #exult
[12:54:48] --> Colourless has joined #exult
[12:54:48] --> cnj has joined #exult
[13:05:15] <-- Colourless has left IRC ("brb")
[13:09:38] --> Colourless has joined #Exult
[13:09:39] --- ChanServ gives channel operator status to Colourless
[13:09:45] <slacked> Colourless,
[13:09:53] <slacked> I was just thinking about you
[13:10:00] <Colourless> yes?
[13:10:07] <slacked> speak of the devil
[13:10:10] <Colourless> i see
[13:10:25] <slacked> so how's work with exult going?
[13:10:44] * slacked trys to reassure that he is a 'nice' stalker...
[13:10:59] <Colourless> jeff would be the person to ask that question too. he's the only one doing anything afaik
[13:11:24] <slacked> okies...
[13:11:39] <slacked> Does that Trent guy still work on exult anymore?
[13:12:51] <Colourless> i don't know how you are talking about
[13:13:41] <slacked> the first guy
[13:13:47] <slacked> on the change log
[13:14:16] <slacked> gar, internet refuses to work.
[13:14:37] <Colourless> you mean Tristan
[13:14:40] <slacked> yah
[13:14:40] <Colourless> aka Nadir
[13:14:43] <slacked> thats the one
[13:14:46] <slacked> hehe
[13:15:18] <slacked> I feel like playing exult 0.2 alpha
[13:15:29] <Colourless> no, he hasn't really been involved at all for a while
[13:15:41] <slacked> just to remember what it was like in the 'good old days'
[13:15:47] <slacked> Oh, I see.
[13:16:38] <slacked> dancer added --enable-mmx and --enable-3dnow. I tried ./configure with those options and it says it can't find the,
[13:18:55] <slacked> is it even programmed yet? or has he just added those incase anyone wants to work on them
[13:30:06] <Darke> I think he was working on the code (there should be a mention in the logs about him having them working, IIRC), don't know if he committed them or not.
[13:30:14] <Darke> Maybe check the changelog?
[13:32:12] <Darke> Or do a grep of the source files for 'MMX' and '3DNOW'?
[13:32:59] <Colourless> there was no changes to the source
[13:33:19] <Colourless> it was just enabled some gcc compiler options or something
[13:33:31] * Darke ahhs. Ok.
[13:33:33] <Colourless> i have no idea what it actually does
[13:35:31] <Darke> Yeah, it just adds two flags (one for each option) to the CXXFLAGS variable.
[13:48:01] <-- slacked has left IRC (Read error: 110 (Connection timed out))
[15:28:22] <-- Colourless has left IRC ("bbl")
[15:41:02] --> Colourless has joined #Exult
[15:41:02] --- ChanServ gives channel operator status to Colourless
[15:47:03] <-- Kirben has left IRC (Read error: 104 (Connection reset by peer))
[17:44:05] <Darke> Hmm... compiling whilst playing exult really gives you that u7-on-a-486/33 feeling.
[17:48:29] <Darke> Yes there is a problem with the 'glowing text' in the spell book. Two problems in fact, and maybe a third. *grin*
[17:49:18] <Darke> Firstly the text for the name isn't spaced out properly. "First" for example lacks a single vertical row of seperating pixels between each letter, they all appear squished up against each other.
[17:49:57] <Darke> Secondly the numbers for the reagents are one 'letter' too far to the left, they're not properly centered in the blank space that that's allocated for them.
[17:50:32] <Darke> (Actually, didn't it max out displaying 99 reagents too, rather then 100? Can't remember, it's been a while. *grin*)
[17:50:44] <Colourless> 99
[17:51:50] <Darke> Also, there's infinity signs in the '0th level' spell section where a number should be. The problem is there's no blank area in the 0th circle spells like there is for the 1-8th level. I'm not sure if it was like that in the original or not, either way it looks 'wrong', IMO.
[17:52:12] <Darke> If it maxes out at 99, it'll look ok. With the 100, the '1' overlaps the 'text' graphic on the left.
[17:53:04] <Colourless> infinity symbol was only supposed to be used in SI with the Ring of Reagents
[17:55:02] <Darke> On a weird note, it looks like with mmx/3dnow exult's just that much less cpu intensive, 10-20% I'm guessing from looking at things.
[17:55:38] <Darke> Cool. Feel like filing a bug report/fixing it? Or shall we let the omnisentient DrCode do it for us? *grin*
[17:55:53] <Colourless> no i don't feel like doing any of those
[17:56:44] * Darke wouldn't have a clue where to look. *grin*
[17:57:49] <Darke> Anyway, I just figured I'd best verify that from the forum since the thread seems to get continually hijacked. *grin*
[18:23:23] * Darke wanders off to get some of this 'sleep' stuff. Night!
[18:23:36] --- Darke is now known as Darke|zzZ
[18:26:14] <Colourless> cya
[18:35:38] <-- Colourless has left IRC ("casts invisibility")
[19:07:39] --> wjp has joined #exult
[19:07:39] --- ChanServ gives channel operator status to wjp
[19:09:23] --> matto|wookin has joined #exult
[19:10:18] <matto|wookin> mmm lunch time
[19:14:41] <wjp> hi :-)
[19:15:04] <matto|wookin> hey wjp
[19:15:32] <wjp> what? no exclamation marks? *sob*
[19:15:35] <wjp> ;-)
[19:15:38] <matto|wookin> WJP!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
[19:15:50] * wjp feels a lot better now :-)
[19:16:13] <matto|wookin> I am having an interesting problem with sockets... I am trying to send huge files over TCP/IP as fast as possible with as large of chunks as possible
[19:16:22] <matto|wookin> and I seem to be hitting some kind of overflow condition that causes problems
[19:17:06] <wjp> can you be more specific?
[19:17:18] <matto|wookin> yes... here's what I'm doing
[19:18:26] <matto|wookin> I am sending like 4,000,000 bytes (this is on LAN) using the 'send' command but I am splitting it up into 16384 chunks.. so I call 'send' 4,000,000/16384 times in a row without checking to see whether the receiver has gotten it... I am just counting on 'send' to block if it can't get the data through fast enough
[19:18:30] <matto|wookin> then on the receive end,
[19:18:52] <matto|wookin> I allocate a big 4 meg chunk of RAM and call 'recv' repeatedly until that 4M chunk of RAM is full
[19:19:15] <matto|wookin> this is a multi-threaded model so I am using blocking I/O
[19:19:35] <matto|wookin> basically I am just writing a file transfer program
[19:19:56] <matto|wookin> the receiving thread then sends a little packet to the sender requesting the next large chunk of data
[19:20:03] <matto|wookin> and the sender begins sending ...
[19:20:22] <matto|wookin> I have it working properly if I have the sender sleep for 5 seconds before sending its next chunk
[19:20:48] <matto|wookin> but if I have no delay, then the sender appears to block forever without sending anything (according to my bandwidth monitor) and the receiver eventually times out and disconnects
[19:21:14] <matto|wookin> in other cases, the receiver gets back a length of 0 from recv which I have been lead to believe always indicates that the remote host disconnected
[19:21:15] <wjp> so it receives the "send me more data" packet and then seems to freeze?
[19:21:19] <matto|wookin> but now I am wondering if that is necessarily true
[19:21:47] <matto|wookin> it receives the 'send me more' packet and appears to freeze inside the 'send' function
[19:21:52] <matto|wookin> I don't think it freezes before that
[19:22:02] <matto|wookin> (because I put in a few print statements hehe)
[19:22:07] <wjp> :-)
[19:22:14] <matto|wookin> like .. "We're about to call send again!" hehe
[19:23:17] <wjp> weird; I can't immediately think of any reason for this
[19:23:36] <wjp> you could strace the sender/receiver to see if that shows anything weird
[19:23:38] <matto|wookin> if it was working, it would be a pretty neat program because it automatically calculates the size of the next chunk of data to request from the sender based on the current transfer rate.. so on a LAN it can really transfer massive amounts of data very quickly.. and on dial-up it should work too
[19:24:43] <matto|wookin> strace...
[19:25:14] * matto|wookin does apt-get install strace
[19:28:41] <matto|wookin> hehe.. this will be difficult since it's multi-threaded
[19:29:05] <wjp> -f should follow forks
[19:29:17] <wjp> but don't know about other types of threads
[19:30:00] <matto|wookin> ok that seems to be working
[19:40:29] <matto|wookin> mmm might've found the problem :)
[19:43:59] <wjp> what was it?
[19:47:19] <matto|wookin> I didn't after all :(
[19:47:26] <matto|wookin> but it appears to be something wrong with windows .. hehe
[19:47:28] <matto|wookin> or Winsock rather
[19:47:56] <wjp> aaah.. you're using windows :-)
[19:48:06] <matto|wookin> one host is windows, the other host is linux
[19:48:16] <matto|wookin> I am sending from the windows to the linux
[19:48:16] <wjp> sender is linux?
[19:48:24] <wjp> hm, k
[19:48:40] <matto|wookin> according to windows, it send 7,960,000 bytes plus the trailing 4-byte CRC32 value
[19:49:02] <matto|wookin> according to linux, it was in the middle of receiving 7,960,000 bytes when it read 0 bytes (disconnect)
[19:49:10] <matto|wookin> it then closed its socket
[19:49:16] <matto|wookin> but windows hasn't detectged any disconnection at all
[19:49:24] <matto|wookin> it's just sitting there ...
[19:49:45] <matto|wookin> I have really no way of knowing where it's stuck...
[19:50:02] <matto|wookin> but since it has finished sending I have to assume it's stuck on a select statement (checking for incoming data)
[19:52:51] <matto|wookin> yesssss... it's stuck on the select!
[20:03:59] <matto|wookin> Ok, I solved it for real this time :)
[20:04:13] <matto|wookin> the problem was that I was telling select to block for 10 seconds
[20:04:22] <matto|wookin> but it was returning instantly that there was no data available
[20:04:27] <matto|wookin> (without waiting for 10 seconds)
[20:04:42] <matto|wookin> so I was assuming that meant that the remote connection had crashed or something, so I was closing the thread
[20:05:00] <matto|wookin> but in actuality apparently Winsock doesn't always honor the timeout variable for select if it knows there can't possibly be any data available
[20:05:01] <wjp> ah, while you should've just tried again later?
[20:05:05] <matto|wookin> (linux might do this too)
[20:05:17] <matto|wookin> yes I should've just tried again later
[20:06:34] <matto|wookin> I was assuming that select would _always_ block for 10 seconds unless there was an 'event' available
[20:06:46] <matto|wookin> because I didn't want to put any SDL_Delay() calls inside my thread
[20:07:01] <matto|wookin> now I will need to do so.. and also will need to keep track of elapsed time..
[21:39:58] --> artaxerxes has joined #exult
[21:39:58] --- ChanServ gives channel operator status to artaxerxes
[21:40:01] <artaxerxes> hi all
[21:40:29] <wjp> hi
[21:40:39] <artaxerxes> I'm popping up just to say that I'm really overloaded at work and it'll get better only in week or two
[21:41:34] <artaxerxes> I've been given a project and I forced my boss to give me freehands to do it. :) Conclusion, I'm learning a lot of very useful stuff like, XML, DTD, XSL and all that jazz! :)
[21:42:14] <artaxerxes> btw, if you see Dominus around, I've sent him an email about XML but got no response yet. It's been about 2 days I think.
[21:46:51] <artaxerxes> anyways.. that's the front news?
[21:56:15] <artaxerxes> I have to leave... see ya later all!
[21:56:25] <-- artaxerxes has left IRC ("using sirc version 2.211+KSIRC/1.1")
[22:28:49] <-- wjp has left IRC ("Zzzz...")
[23:15:35] --> Dark-Star has joined #exult
[23:25:26] <matto|wookin> artaxerxes!!!!
[23:25:28] <matto|wookin> oops.. too late