code logs -> 2008 -> Sat, 19 Apr 2008< code.20080418.log - code.20080420.log >
--- Log opened Sat Apr 19 00:00:10 2008
00:04 Reiver [~reaverta@Admin.Nightstar.Net] has quit [Quit: Off to wellington. Back in 4 days or so.]
00:10 DiceBot [~Reiver@Nightstar-8904.xdsl.xnet.co.nz] has quit [Quit: DiceBot: based on AnnoDomini's KarmaBot.]
00:21 * MyCatVerbs dances along to the sounds of drunk Irish people.
00:23
< Shoukanjuu>
Hmm...It's a shame there aren't drivers for the Tarantula on OS X >.>
00:38 * MyCatVerbs debates the merits of dancing nekkid instead.
00:42 * McMartin tires of writing XML output code, goes to make dinner.
04:01 Vornotron [~vorn@Admin.Nightstar.Net] has joined #code
04:02 Vornicus [~vorn@ServicesOp.Nightstar.Net] has quit [Ping Timeout]
04:02 Vornotron is now known as Vornicus
05:39 Vornicus is now known as Vornicus-Latens
07:38
<@C_tiger>
Question: how much more expensive is it to store variables in a file on the disk vs storing them in memory?
07:39
<@jerith>
At least an order of magnitude.
07:39
<@jerith>
Although disk-cacheing can bring that down.
07:40
<@jerith>
Why would you want to do that?
07:40 mode/#code [+U] by jerith
07:40
<@jerith>
(Because the topic said it was set.)
07:40
<@C_tiger>
Well to protect against the contingency of my program going down and then forgetting everything.
07:41
<@C_tiger>
There's a largish hash that I would like to keep. I could tie it to a file or I could not.
07:42
<@C_tiger>
but I access it often.
07:43
<@jerith>
Write it to the file whenever it changes, read from memory.
07:44
<@jerith>
That makes writes more expensive.
07:44
<@C_tiger>
Yeah, I update the variable often.
07:48
<@C_tiger>
It's also a hash of hashes and I'm not sure how to do that with tie.
07:54
<@C_tiger>
I'm trying to figure out if I can just keep data I can't get back easily.
09:22 Raif [~corvusign@Nightstar-25074.hsd1.wa.comcast.net] has joined #Code
09:22 mode/#code [+o Raif] by ChanServ
10:31 McMartin [~mcmartin@Nightstar-18930.dsl.pltn13.sbcglobal.net] has quit [Quit: maintenance shutdown]
10:47 McMartin [~mcmartin@Nightstar-18930.dsl.pltn13.sbcglobal.net] has joined #code
10:48 mode/#code [+o McMartin] by ChanServ
11:32
<@McMartin>
Man. It has been a long time since I've had a bug caused by neglecting to flush streams on program exit.
13:31 You're now known as TheWatcher[afk]
18:19 Vornicus-Latens is now known as Vornicus
18:30 You're now known as TheWatcher
19:38 Zemyla [~Zemyla@168.53.171.ns-21272] has joined #code
19:39
< Zemyla>
How many of you know what fixed-point arithmetic is?
19:44 * Vornicus does!
19:44
< Zemyla>
Because I'm trying to write a library for C++ that basically automates writing code with fixed-point math in it.
19:45
< Vornicus>
Zemyla: okay, that's not too horribly hard.
19:46
< Zemyla>
It is if you're trying to make sure everything fits together, works optimally, and uses templates.
19:51
< Zemyla>
Basically, what I'm trying to do is make sure that everything retains as much precision as possible. Also, using templates will allow the compiler to use constant folding, dead code removal, and inlining to create near-optimal code.
19:52
< Vornicus>
okay. so far I don't see anything out of the ordinary
19:53
< Zemyla>
Well, I have to define helper classes for the operations that carry precision with them.
19:53
< Vornicus>
Right
19:53
< Vornicus>
I really don't see why you need any help with this; what you've said so far makes it pretty clear you know how it all works.
19:54
< Zemyla>
I know, it's just a lot of code writing.
19:54
< Zemyla>
Did you know that C++ templates are Turing-complete?
19:55
< Vornicus>
Yes
19:58
< Zemyla>
Also: Fucking Google Documents wants everything to be in Verdana 10 pt, and resists strenuously any attempt to change it to another font. >_<
20:00
< Vornicus>
I'm sorry.
20:00
< Zemyla>
It's not your fault, it's just irritating.
20:01
< Shoukanjuu>
Why are you doing this in google documents? o.o;
20:03
< Zemyla>
Because I'm on a library computer.
20:05
< Zemyla>
This way, stuff is saved.
20:24
< Zemyla>
Each floating-point-style operation on a pair of fixed-point numbers will instantiate a class that, when passed the desired type of the target, will produce an outpot un that desired type, retaining as much precision of the two input types as possible.
20:28
< Zemyla>
Then, when the compiler compiles it, hopefully all the inlining will turn it into just a series of arithmetic operators.
20:31
< Zemyla>
Hmm, I need to implement pre- and post- ++ and --, unary +, -, !, and integer and floating-point conversion, binary +, -, *, /, comparison operators, and ternary ?:.
20:32
< Zemyla>
The reason I'm implementing ?: is because you can do ((x > y) ? a : b) = c, and I need to make sure that the proper conversion is done in each case.
20:33
< Zemyla>
Wait, can I overload the ternary operator?
20:33
< Zemyla>
No, I cannot. Damn damn damn!
21:16
< Zemyla>
Anyways, I have to go, bye everyone.
21:16 Zemyla [~Zemyla@168.53.171.ns-21272] has quit [Quit: http://irc.netsplit.de/]
22:04 Kazriko [~kaz@host-72-174-26-165.gdj-co.client.bresnan.net] has quit [Ping Timeout]
22:09 Kazriko [~kaz@Nightstar-26352.gdj-co.client.bresnan.net] has joined #code
22:09 mode/#code [+o Kazriko] by ChanServ
--- Log closed Sun Apr 20 00:00:25 2008
code logs -> 2008 -> Sat, 19 Apr 2008< code.20080418.log - code.20080420.log >