Perl 6 - the future is here, just unevenly distributed

IRC log for #rosettacode, 2012-02-28

| Channels | #rosettacode index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:06 mikemol sorear: I don't know. I haven't/hadn't set up a good way to set up a good way to track it.
00:11 mwn3d_phone Yeah I wonder if it takes up a good portion of the bandwidth limit
00:38 mikemol Certainly not
00:38 mikemol I've never had a stream go to the server faster than 100B/s, and never more than three simultaneous streams.
00:40 mikemol Someone know the glibc/printf code for printing a 64-bit unsigned int in decimal?
00:40 Coderjoe %lu iirc (for gcc/glibc)
00:40 mikemol Ah, nm. Found the length modifier.
00:51 mikemol Ok, I now have a variant buffering up to 8MB of entropy which also tracks how many bytes it's read in and out over the life of the program. Also, it caps its input and output at 128B/s instead of 64B/s.
01:05 Coderjoe http://channel9.msdn.com/Events/Goi​ngNative/GoingNative-2012/Keynote-B​jarne-Stroustrup-Cpp11-Style/player
01:05 fedaykin "Day 1 Keynote - Bjarne Stroustrup C 11 Style GoingNative 2012 Channel 9" http://rldn.net/sBfv
01:06 mwn3d_phone1 joined #rosettacode
01:06 Coderjoe 128B is 1/4 of your maximum kernel entropy pool size
01:07 * mikemol nods
01:07 mikemol It's also the distance between the upper and lower thresholds for operation.
01:07 Hypftier joined #rosettacode
01:08 Coderjoe so it will yoyo
01:08 mikemol No; there's another 128B between the upper threshold and the kernel's internal cap.
01:10 mikemol Though I should probably adjust one of those comparisons to not include "or equal", or there is a chance of yoyo if it hits 2048/3072 exactly.
01:10 sorear mikemol: why not just hack the kernel to boost the pool to 8Mo or so?
01:10 sorear does prgmr not let you run custom domU kernels?
01:11 Hypftier joined #rosettacode
01:11 mikemol sorear: A variety of justifications. 1) I'm not comfortable doing this in kernel space if I don't need to. 2) I'm not comfortable kernel hacking on Debian. 3) I don't want to risk running afoul of some assumption in the kernel codebase where it's assumed the entropy pool will be 4096 bits.
01:12 mikemol It's simply safer to do it in user space. Also, less down time; I can kill and spawn as many of these userland processes as I want.
01:15 mikemol Eventually, I think I want to play with some adaptive behaviors in the entropy capacitor. Such as adjusting its inflow/outflow limits and its poll frequency in response to starvation/overflow events. And possibly backing the buffer with a file, and allowing it to grow.
01:15 mikemol It's gone beyond a problem needing to be solved, and into "hey, this is interesting" territory.
01:18 mikemol I should probably also add "respond to SIGUSR1 as an instruction to stop filling the buffer"; right now, I have no good way of killing the program without losing the buffer it's saved up. It'd be nice to be able to spawn a new buffer, have it start filling, and tell the other to only drain.
01:22 Coderjoe_ joined #rosettacode
01:22 mikemol K, right now, I have three entropy streams heading out from home. One is pushing 141B/s, one is pushing 87B/s, one is pushing 66B/s. Hm. I can probably kill the two slower feeds; 141B/s is faster than the buffer will charge, and anything beyond that which isn't actively being consumed is being wasted.
01:22 Hypftier joined #rosettacode
01:22 mikemol Huh. If I had to guess, I'd say the server's active consumption rate is hovering right around 30-40B/s right now, based on the buffer not filling as quickly as it could.
01:23 Coderjoe_ oooooh
01:24 Coderjoe_ c++11 "move constructor" looks neat
01:24 mikemol Yeah, been looking forward to that.
01:24 mikemol Coderjoe_: Move semantics are already in VS2010, btw.
01:24 mikemol Well, IIRC, anyway.
02:20 maxsilver joined #rosettacode
02:21 eMBee what can that move constructor do?
02:24 mikemol eMBee: Move state from the source object to the constructed object without actually copying data.
03:08 maustin joined #rosettacode
03:16 Coderjoe joined #rosettacode
04:33 Coderjoe eMBee: quite useful for returning an object from a function
04:34 saschakb_ joined #rosettacode
04:35 Coderjoe for example, if you had a Matrix class with a move ctor and a multiplication operator, you could do: /* create instances of Matrix named A and B */ Matrix prod = A * B;
04:36 Coderjoe and rather than making a copy of the returned stack object with the copy ctor, prod IS the returned object
04:41 * eMBee doesn't know c++, so you couldn't just return prod;?
05:07 ttmrichter joined #rosettacode
05:29 ko1 joined #rosettacode
06:22 Coderjoe eMBee: depending on how the function is defined, that would return a reference (likely to a stack object), a pointer, or make a copy
06:22 Coderjoe none of which do the move
07:06 ko1 joined #rosettacode
07:11 eMBee ok, so howisa move different from areference? i mean, is a reference not enough?
07:13 sorear eMBee: you're talking like someone who uses languages with garbage collection
07:13 sorear in C++, returning references to local variables is simply not an option.
07:14 eMBee that observation is correct :-)
07:16 eMBee will c++ prevent that or will trying  that have unknown consequences?
07:16 sorear the usual, nasal demons
07:17 Hypftier At least in C you have problems if you return a pointer to a local, as it points somewhere on the stack and thus doesn't live past RET – is that a similar problem to C++ with references?
07:18 sorear C++ references are pointers
07:18 Hypftier only that you can assign them only once ;)
07:19 sorear references are syntax sugar which auto-dereferences on all access except the initializer
07:52 mischi joined #rosettacode
08:21 mwn3d_phone1 joined #rosettacode
09:02 ttmrichter joined #rosettacode
09:30 kpreid joined #rosettacode
10:16 ttmrichter joined #rosettacode
10:23 * mikemol whistles
10:25 mikemol So I'm still feeding about 75B/s from home. The buffer is at around 4MB, but its keeps stalling for one second a couple times a minute.
10:26 mikemol Er. I'm feeding at 59B/s + 75B/s.
10:26 * mikemol looks forward to that entropy key.
10:33 mwn3d_phone joined #rosettacode
12:51 mwn3d_phone1 joined #rosettacode
14:01 robbrit joined #rosettacode
15:12 maxsilver_ joined #rosettacode
15:14 maxsilver joined #rosettacode
15:23 Coderjoe joined #rosettacode
15:59 mischi joined #rosettacode
16:01 kpreid joined #rosettacode
16:27 mwn3d_phone joined #rosettacode
16:27 mwn3d_phone mikemol: what did the user "she tekierso" just do?
16:28 mwn3d_phone [[Special:Log/newusers]]
16:28 fedaykin http://rosettacode.org/wiki/Special:Log/newusers
16:28 mikemol Spam, IIRC.
16:28 mikemol I haven't had to block anyone for anything else in ages.
16:28 mwn3d_phone No one blocked them yet
16:29 mwn3d_phone But I've never seen a user create a user before
16:29 mikemol ah
16:29 mikemol Sure, users can create users.
18:05 kpreid joined #rosettacode
19:07 mwn3d_phone joined #rosettacode
19:07 mwn3d_phone Uh oh...
19:07 mwn3d_phone The diff is too large to get an undo link
19:13 mikemol ?
19:13 mwn3d_phone Check the recent changes
19:13 mwn3d_phone Big mistake on fibonacci sequence
19:14 mikemol haha
19:14 mwn3d_phone Oh we can rollback
19:14 mikemol 112,364 bytes.
19:14 mikemol hold off
19:14 mikemol Don't roll back.
19:14 mwn3d_phone K
19:15 mikemol Trying an undo. It's spinning CPU, we'll see what happens.
19:15 mikemol k. Failed.
19:15 mikemol Checking the logs to see what resource what exceeded.
19:16 mikemol Max execution time failure.
19:16 mikemol I'll tweak it.
19:17 mikemol Trying again
19:26 mikemol Copied his code to his talk page.
19:26 mikemol Rolled back his changes.
19:27 mikemol 120 sec max execution time isn't enough, and I'm not bumping it any higher than that.
20:59 kpreid joined #rosettacode
21:04 robbrit left #rosettacode
21:18 mwn3d_phone joined #rosettacode
21:18 ko1 joined #rosettacode
21:18 Hypftier joined #rosettacode
21:18 dagnyscott joined #rosettacode
21:18 slavik1 joined #rosettacode
21:18 p6eval joined #rosettacode
21:18 mbishop joined #rosettacode
21:18 mikemol joined #rosettacode
21:18 eMBee joined #rosettacode
21:18 Util joined #rosettacode
21:18 dom96 joined #rosettacode
21:18 opticron joined #rosettacode
21:18 FireFly joined #rosettacode
21:18 ozzloy joined #rosettacode
21:48 lambdabot joined #rosettacode
22:08 mwn3d_phone joined #rosettacode
22:17 mischi joined #rosettacode
22:23 saschakb joined #rosettacode
22:41 kpreid joined #rosettacode
23:53 kpreid joined #rosettacode

| Channels | #rosettacode index | Today | | Search | Google Search | Plain-Text | summary