Perl 6 - the future is here, just unevenly distributed

IRC log for #rosettacode, 2011-07-27

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

All times shown according to UTC.

Time Nick Message
00:02 rodt mike thanks, yeah its a good idea :D still alot of low lvl rubbish to wade thro at the mo tho... sorry late delay lol
00:13 BenBE Hypftier You mean let Wikipedia have a slider for "My Grandma" to "Prodigy"???
00:14 Hypftier maybe ;)
00:14 rodt great idea :)
00:15 Hypftier The point is, orphaned low-quality articles that someone wrote once aren't going to hurt anyone. Storage is dirt cheap these days.
00:15 Hypftier But many things get deleted due to the notability guidelines just because it's against the rules but would still be something you might want to look up
00:16 BenBE Hypftier There was a notability discussion going on already in 2009 and at the end of 2009 on the CCC's 25C3 congress they had a vivid discussion on that topic. If you feel like watching that talk: You'll find one of the audience questions being asked by my humble person ;-)
00:16 Hypftier This includes many things that are currently researched and have peer-reviewed papers on, but no newspaper article, because somehow journals are less notable than newspapers.
00:16 Hypftier BenBE: I was there and watched the stream ... it's been a while, though :)
00:16 BenBE Basic line on my PoV: Include things and let the reader decide ...
00:16 Hypftier And yes, the German WP was the main source of that idea.
00:17 rodt but yeah automatic natural language manipulation is ai-complete or something
00:17 Hypftier I later noticed that en.wp isn't much different; it just happens to include lots of stuff that wouldn't ever be on de.wp (e.g. fictional characters or places)
00:18 BenBE Hypftier EnWiki backup is 273GiB (last checked March 2010, when they finally managed to create one again); dewiki follows with ~70GiB; both figures for the tar.gz versions of the files; 7z being much better compression-wise.
00:18 * Hypftier won't do it automated. It still requires a fairly intricate model (which I hope to get around specifying it this week) and plenty of work on the author's behalf. I cannot and won't do automatic bridge insertion to maintain coherence ;)
00:19 Hypftier BenBE: I rarely check dewiki nowadays, except for local topics I'm fairly sure to be represented better in dewiki :-)
00:19 rodt authors are lazy tho ?
00:19 Hypftier yes, but they are better than computers at such things :P
00:19 rodt ;)
00:21 Hypftier Also, similar things are already necessary in e-learning and intelligent tutoring systems with not exactly less work. It might be extendable to groups of authors, though, but I have to hand in the thesis in two months, so I should restrict myself a little. A diploma is no place for PhD-level research :P
00:22 rodt what are you doing it on ?
00:22 rodt sorry ive prolly missed the plot ?
00:22 Hypftier *nods* you did. Just scroll up ;)
00:22 rodt :)
00:23 rodt oh nice :)
00:25 rodt really difficult tho to build a model of the lang from teh small descriptions
00:26 rodt maybe they are tabulised somewhere i havent seen yet on RC ?
00:27 rodt Ada is a structured, statically typed imperative computer programming language initially standardized by ANSI in 1983 and by ISO in 1987. It is called as Ada 83. ....etc...
00:27 Hypftier Yeah, RC is a bit short on text in general for that to work; and given how bad an author I am I doubt I could make something good from that :)
00:28 rodt you might get away with pulling in another resource tho, like model(wiki/ADA) or some such
00:28 rodt prefferable tabulations
00:29 rodt column labeled tables*
00:29 Hypftier Well, for now it's bedtime. 2:29 here :)
00:29 rodt sweet dreams :)
00:49 mikemol TimToady: Ooh. You're right; robots.txt disallows /mw/ and /w/
00:49 mikemol I'll need to fix that.
00:52 FireFly left #rosettacode
01:14 mikemol Fixed robots.txt to allow pulling from /{w|mw}/images/, /{w|mw}/skins/ and /{w|mw}/title.png
01:17 mwn3d_phone Daw....got locked out temporarily cause I played a video game where caps lock is the sprint toggle and I forgot to take it off :/
01:17 mwn3d_phone How long does that last?
01:22 mikemol mwn3d_phone: wiki locked you out?
01:22 mikemol I didn't know it did that.
01:26 mikemol mwn3d_phone: I'll guess it'll only lock you out for fifteen minutes. I'm not even coming up with the right keywords to find documentation of the behavior.
01:27 mwn3d_phone Too many login attempts in a small timespan
01:27 mwn3d_phone I'll be ok
01:30 mwn3d_phone And I'm back
01:36 kpreid left #rosettacode
01:58 mikemol mwn3d_phone: Incidentally, it looks like the Widgets extension is installed.
01:58 mwn3d_phone What's that do?
02:07 kpreid joined #rosettacode
03:34 Coderjoe left #rosettacode
03:34 Coderjoe_ joined #rosettacode
03:36 BenBE left #rosettacode
04:10 Coderjoe_ "This sample implements a Hidden Markov Model in OpenCL for the GPU."
06:46 damagednoob joined #rosettacode
07:40 Hypftier TimToady: HTTrack worked fine in mirroring the site, by the way; I now have a 543 MiB dump that has CSS and images
08:18 damagednoob left #rosettacode
08:55 rodt left #rosettacode
10:39 BenBE joined #rosettacode
11:17 FireFly joined #rosettacode
12:05 mwn3d_phone left #rosettacode
12:09 BenBE left #rosettacode
12:14 BenBE joined #rosettacode
12:15 mwn3d_phone joined #rosettacode
12:44 mikemol Hypftier: Tarball that and pass it my way?
13:08 Hypftier mikemol: http://hypftier.de/temp/RC.7z
13:08 Hypftier 258 MiB (of which most probably falls in a few .tar.bz2 archives)
13:22 Hypftier (ok, if I remove those, around 60 MiB remain ... should I re-upload?)
13:29 dagnyscott1 joined #rosettacode
13:37 mikemol Hypftier: Up to you; I'm pulling from the rc server, so it's no skin off my teeth.
13:37 mikemol Currently pulling at about 1M/s, eta 4m
13:38 Hypftier I had only 20 to 40 KiB/s ... couldn't get httrack to use more connections ;)
13:39 Hypftier Ok, RC2.7z, same location. 61 MiB
13:39 mikemol Probably just as well. If it was CPU limited serverside (and it probably was), then more connections wouldn't have helped.
13:39 mikemol Well, except for the static data under /skins/ and /images/
13:42 mikemol Ok, rehosted at http://rosettacode.org/resource​s/2011-07-27-rosettacode.org.7z
13:42 * mikemol eyes fedaykin
13:43 fedaykin http://rldn.net/s44H
13:46 opticron heh
13:46 opticron yeah...I still haven't fixed that
15:09 TimToady the new snapshot looks pretty good, thanks mikemol++ and Hypftier++
15:10 TimToady (the only glitch is that I have to tell firefox to work offline, or it still tries to find the website)
15:10 mikemol Any idea why?
15:12 Hypftier There are still a few links to the site in there (edit, for example), maybe there are also a few resources linked from there :|
15:13 Hypftier The search provider still points to rc.org, for example.
15:13 Hypftier oh damn ... and stylesheets too :O
15:13 * Hypftier goes hiding in a corner
15:14 TimToady but the pages I'm interested in come up nicely when I do the offline thing
15:14 TimToady so I'm not too worried
15:14 Hypftier cache, probably
15:14 TimToady mebbe
15:14 Hypftier hit Ctrl+F5 and the style disappears
15:14 Hypftier at least here :/
15:14 mikemol If we can get a script put together to get this cleaned up in an automated fashion, and get a reasonable static snapshot together, I'd like to go a bit further and automatically publish packages people can install on their local systems.
15:15 TimToady it wouldn't bother me to continue to improve it :)
15:15 mikemol It'd need to get divided up, though, and the images would need to be splittable to their own package.
15:16 mikemol But it's something I'd love to be able to do; kinda like how LinuxGazette has been available as a Debian package for ten years...
15:16 TimToady as you know, it's a recurring issue for me :)
15:16 * mikemol nods
15:18 Hypftier serverdump + local webserver/ :D
15:18 TimToady if that's what it takes...
15:19 * Hypftier tries tweaking the mirror settings. Maybe it didn't like that half of the CSS had a ?270 at the end of the URI and the other half went through index.php
15:19 TimToady don't wanna become a database expert though...
15:19 mikemol I can't push that kind of dump until I've got an automagic way of stripping private details out of the database.
15:20 mikemol TimToady: It's probable that it can eventually be pushed in a fairly self-contained form; MediaWiki supports SQlite as a backend, so if I can eventually convert the MySQL database to SQLite as a stage in a regular publishing cycle, the sqlite bit obviates the need to run a database server.
15:21 mikemol But we're talking crazy now, spending time I don't have at the moment.
15:21 * Hypftier taxes the server again ;)
15:40 Hypftier BenBE: [[Tokenize_a_string#Batch_File]] has a rendering problem, apparently: http://hypftier.de/temp/2011-07-27_174031.png
15:40 fedaykin http://rosettacode.org/wiki/​Tokenize_a_string#Batch_File
15:49 mwn3d_phone mikemol: I don't know when a MW upgrade can be done but I'm not sure if it'll have any impact on this process. I don't know how much this stuff is getting into the DB and how much an upgrade would change it.
15:49 mwn3d_phone Now that I think about it though you guys are all talking in URLs so it probably won't impact it at all
15:49 realazthat_ is now known as trolzies
15:50 mikemol The real reason I haven't done a MW upgrade lately is because my automatic backups have been down again, manual backups are time consuming, upgrades are time consuming, and that makes (time consuming)^2.
15:50 mikemol I'm regularly busier than I've ever been in my life, so figuring out where to fit these things is difficult.
15:51 mwn3d_phone Heh you don't need to give me a reason for not upgrading. I trust that you have good reasons.
15:51 mikemol I wouldn't count 'nonfunctional backups' as a good reason for *anything*...
15:51 mwn3d_phone Just making sure you're thinking about things that might impact this offline copy process
15:52 mikemol I was *pissed* when I discovered that mysqldump's "xml" output can't be used for database restoration or, really, anything else.
15:52 mikemol It just looks pretty, if you find XML pretty.
15:52 mikemol Well, badly-formed XML.
15:53 mikemol I was going to use that and xmlstarlet to strip private details from the database prior to regular publishing. :(
17:05 dagnyscott1 left #rosettacode
17:07 dagnyscott1 joined #rosettacode
17:50 kpreid left #rosettacode
18:12 mwn3d_phone1 joined #rosettacode
18:13 mwn3d_phone left #rosettacode
18:16 kpreid joined #rosettacode
18:26 mwn3d_phone1 is now known as mwn3d_phone
18:48 rodt joined #rosettacode
20:06 mwn3d_phone Edits are really slow for me
20:06 mwn3d_phone Might just be temporary...the last one was quicker
20:07 * Hypftier whistles innocently
20:08 mwn3d_phone -_-
20:08 mwn3d_phone As long as we know what's going on its ok
20:12 mikemol I've got Hypftier set up with an account on the server, so he can run (and tune) httrack locally. Had a load spike for a bit there.
20:13 Hypftier it's back to 1 now, though.
20:13 mikemol Most users won't be affected, thanks to the squid caching, but things which fall through to poking php will be.
20:15 mikemol I've just disabled hit counting on the server.
20:15 mikemol This means caching will be invalidated quickly for  a bit, but will get more stable after a while.
21:00 mwn3d_phone Is that forever or just while work is being done?
21:40 mikemol Probably forever.
21:56 BenBE DOS/Batch problem should be fixed in a moment; just uploading the fixed language file.
21:59 BenBE The double colouring (in the two lines at the bottom of the code sample) are desired behaviour. Standard Color Scheme treats them as one color.
21:59 Hypftier I meant primarily the HTML bleeding through ;)
22:04 BenBE Yeah, I noticed. There was some ... well rerendering of Regexps ...
22:07 BenBE Hope the fixed highlighting is accepted as an improvement ;-)
22:31 BenBE Could someone have a short look at http://rosettacode.org/wiki/Tokenize_a_string#C (second solution)? Hope this is okay.
22:31 fedaykin "Tokenize a string - Rosetta Code"
22:32 BenBE (i.e. the second solution for C)
22:47 rodt temporaryly
22:48 BenBE What about it?
22:49 rodt its spelt wrong
22:49 rodt the string to operate on to be writeable. << this is confusing as well
22:49 rodt the input string to be writeable, maybe better ?
22:49 Hypftier rodt: correct it, then. I guess, BenBE was aiming fo the highlighting, not the content.
22:49 BenBE Okay. Could you fix those two?
22:50 rodt oh writeable = writable as well
22:50 rodt the highlighting works ?
22:50 BenBE Hypftier Nope; just added the second solution for C and wanted someone to have a look at it (didn't test it; just wrote it from brain to wiki)
22:51 BenBE The highlighting works ;-)
22:51 rodt ah it looks ok to me, but i dont use c for awhile
22:51 BenBE Well, if you look at the history you'll notice I had a buffer overrun in it at first ;-)
22:51 BenBE It read one byte behind the end of the string, which IS bad ;-)
22:52 rodt lol
22:52 rodt i never spot that.. its easier getting a overun ;)
22:53 rodt i would edit it but i cant spell either, and my grammer is terrible, i prolly make it worse :P
22:53 BenBE Suppose you run the tokenize function with a callback the itself runs the tokenize function (but for another separator). If you read one char behind the end of the string in the outer callback you'd get wrong results for the inner.
22:53 rodt :S
22:53 BenBE Like A+B+C,B+D,E+F,H+E+L+L+O,...
22:54 rodt it took me three days to find my recent error, i was dropping all the zeros in dropleadzeros() as opposed to just hte first ones :S
22:54 BenBE The function with the buffer overrun would return A,B,C,B,D,E,F,... for the first inner callback, but actually should have only A,B,C.
22:55 * BenBE regularly works on code audits and security stuff ...
22:55 rodt ah
22:56 BenBE I was kind and declared the callback type using typedef. Otherwise the declaration of tokenize would have been kinda funny ;-)
22:56 rodt i find that stuff interesting dont dont like bothering with it... so i prefer langs where it cant happen etc...
22:57 BenBE void tokenize(char *s, char delim, (void *cb)(const char *s)) { ...
22:57 rodt yeah
22:58 BenBE But best thing still is declaring arrays of function pointers ;-)
22:58 rodt blah
22:58 rodt cant get that concept into my language :(
22:59 rodt its terribly annoying lol
22:59 BenBE Function Pointers or Arrays?
22:59 rodt both really, but im not bothered about arrays
23:00 rodt its stream based and i can arbitarily package teh data chunks up
23:00 rodt so arrays are sort of there in a wiered way
23:01 rodt i dunno how to put a graph in there tho :(
23:01 rodt having infinite numbers only makes me feel a little bit better lol
23:02 rodt well it used to , but arithemetic in general is broken... :S
23:02 rodt could fix with function pointers (or equiv in lang) tho :)
23:03 BenBE Callbacks are a kinda nice concept and simplify many things if you need some operation performed (like searching substrings) and work on the intermediate results.
23:04 BenBE Using a callback here saves you the trouble of copying memory around ;-)
23:04 BenBE rodt Could you fix the things you mentioned?
23:04 rodt if you really want me to ?
23:05 BenBE I'd be grateful to you.
23:06 rodt er writable seems ok, it must be a americanism, ill leave that
23:08 rodt ok done
23:10 rodt i made a filter "function" that "accepts" another function
23:10 rodt its just two functions in a loop
23:11 rodt the filter gives the other thing one byte at a time, and the other thing can give it back or give something else back or just throw it away
23:12 rodt the filter then emits the result, once the otherthing says its handled all the bytes
23:16 rodt its sort of inherent with the paradigm, but passing around functions to functions is completely alien to it
23:38 BenBE I hate people that don't have their mail servers under control: http://pastebin.com/Nn1ck8L3
23:39 fedaykin "Folgender Client versuchte alle 60 Sekunden eine Mail zuzustellen: Jul 28 01: - Pastebin.com"
23:39 mwn3d_phone left #rosettacode
23:39 BenBE (The blocking is only calling one shell script with the IP to block and a second to reload the zone file into Bind and Postfix)
23:41 rodt hmm dunno whats wrong with that
23:42 rodt mine did that and then another system checks the block list, for patterns /etc...
23:47 mwn3d_phone joined #rosettacode
23:51 BenBE Have a close look at the sender mail address and the server name advertised in the helo command: The mail is from "server02.coast._local_ and originates from postmaster@_cost.local_
23:53 BenBE I have a local DNSRBL auto-parsed from my server logfiles: If my server rejects a mail for domain forgery it adds that IP automatically to the list of banned IPs.
23:53 rodt ah
23:53 BenBE It's kinda neet Bash magic ;-) Like "generate a Zone File with pure shell scripts ;-))
23:53 rodt cool :)

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