| --- Log opened Mon May 21 00:00:10 2012 |
| --- Day changed Mon May 21 2012 |
| 00:00 | | You're now known as TheWatcher[T-2] |
| 00:03 | | You're now known as TheWatcher[zZzZ] |
| 00:22 | | himi [fow035@Nightstar-5d05bada.internode.on.net] has quit [Ping timeout: 121 seconds] |
| 00:54 | | RichyB [MyCatVerbs@Nightstar-86656b6c.cable.virginmedia.com] has quit [Ping timeout: 121 seconds] |
| 01:25 | <&McMartin> | https://gist.github.com/2760021 <-- love the 20+-binding let clause for doing the year-end computation |
| 01:33 | <&jerith> | What are you rewriting? |
| 01:37 | <&McMartin> | This: http://www.atariarchives.org/basicgames/showpage.php?page=96 |
| 01:39 | <&jerith> | Ah. |
| 01:42 | | himi [fow035@D741F1.243F35.CADC30.81D435] has joined #code |
| 01:42 | | mode/#code [+o himi] by ChanServ |
| 02:07 | | Kindamoody[zZz] is now known as Kindamoody |
| 02:31 | | Attilla [Obsolete@Nightstar-3f363270.as43234.net] has quit [Ping timeout: 121 seconds] |
| 02:37 | <~Vornicus> | Nurbs are more expensive. Unlike b-splines, however, they cannot be accurately conve3rted to beziers. |
| 02:43 | <~Vornicus> | (an order p b-spline with k control points can be converted to iirc k-p bezier splines of order p. |
| 02:43 | <~Vornicus> | (nurbs on the other hand can accurately model all the quadratics including circles, where b-splines and beziers cannot) |
| 02:54 | | Vash [Vash@Nightstar-241cb5d4.wlfrct.sbcglobal.net] has joined #code |
| 02:54 | | mode/#code [+o Vash] by ChanServ |
| 03:01 | <~Vornicus> | Also if you're already working in a projective space (which if you're passing vector4's, you are), then the last d divides are "free" |
| 03:15 | | * Vornicus eyes visio |
| 03:15 | <~Vornicus> | okay so it says this is a nurbs |
| 03:16 | <~Vornicus> | but a nurbs, like a b-spline, needs more knots than control points, but there are clearly the same number on both here. |
| 03:30 | <~Vornicus> | oh oh oh. clamped, don't need to worry about the last knot values because of that. |
| 03:58 | | jeroid [jerith@83BF1D.4E7445.8DFF6A.5145AD] has joined #code |
| 04:03 | | Derakon is now known as Derakon[AFK] |
| 04:07 | | jeroid [jerith@83BF1D.4E7445.8DFF6A.5145AD] has quit [Ping timeout: 121 seconds] |
| 04:11 | | jeroid [jerith@83BF1D.4E7445.8DFF6A.5145AD] has joined #code |
| 04:26 | | jeroid [jerith@83BF1D.4E7445.8DFF6A.5145AD] has quit [[NS] Quit: Bye] |
| 05:05 | < celticminstrel> | I think I like how XCode can detect the proper name to use when I typo. |
| 05:09 | | Derakon[AFK] is now known as Derakon |
| 05:16 | | celticminstrel [celticminst@Nightstar-5d22ab1d.cable.rogers.com] has quit [[NS] Quit: And lo! The computer falls into a deep sleep, to awake again some other day!] |
| 06:03 | | iospace is now known as iospacedout |
| 06:10 | | Derakon is now known as Derakon[AFK] |
| 06:22 | | Rhamphoryncus [rhamph@Nightstar-5697f7e2.abhsia.telus.net] has joined #code |
| 06:22 | | You're now known as TheWatcher |
| 06:30 | | Kindamoody is now known as Kindamoody|P2 |
| 06:44 | | Vash [Vash@Nightstar-241cb5d4.wlfrct.sbcglobal.net] has quit [[NS] Quit: I lovecraft Vorn!] |
| 06:47 | | Noah [nbarr@Nightstar-9e797a1c.pools.spcsdns.net] has left #code ["Leaving"] |
| 06:47 | | Noah [nbarr@Nightstar-9e797a1c.pools.spcsdns.net] has joined #code |
| 06:50 | | eckse [eckse@Nightstar-44362701.dsl.sentex.ca] has joined #code |
| 06:50 | | mode/#code [+o eckse] by ChanServ |
| 06:51 | | eckse_ [eckse@Nightstar-44362701.dsl.sentex.ca] has quit [Ping timeout: 121 seconds] |
| 07:16 | | eckse [eckse@Nightstar-44362701.dsl.sentex.ca] has quit [Client closed the connection] |
| 07:35 | | Reiver [reiverta@5B433A.3CF6C7.B34AFB.F361B9] has joined #code |
| 07:38 | < Rhamphoryncus> | *twitch* |
| 07:39 | < Rhamphoryncus> | auto_ptr is described as a limited garbage collection facility.. all it does is extend your stack-based destructor behaviour on to a single object on the heap |
| 07:40 | < Rhamphoryncus> | Meaning it behaves exactly like a normal stack-based object. It just happens to be dynamically allocated on the heap |
| 07:58 | | himi [fow035@D741F1.243F35.CADC30.81D435] has quit [Ping timeout: 121 seconds] |
| 08:13 | < Rhamphoryncus> | "Ting also compared the human time cost vs. the computer time cost and revealed that the image of the VW cost more than its street value." |
| 08:13 | < Rhamphoryncus> | XD |
| 08:13 | < Rhamphoryncus> | That's *the* VW Bug model from 1972 |
| 08:14 | < Rhamphoryncus> | Which is notable for being the birth of Phong Shading |
| 08:17 | | RichyB [MyCatVerbs@Nightstar-86656b6c.cable.virginmedia.com] has joined #code |
| 08:21 | | Kindamoody|P2 is now known as Kindamoody |
| 08:43 | < Rhamphoryncus> | "FreeType can spit out Bezier glyph boundaries and kerning information" that would be very useful for rendering fonts in opengl, yes |
| 08:52 | < Rhamphoryncus> | but now that I read more the first answer says otherwise: http://stackoverflow.com/questions/5262951/what-is-state-of-the-art-for-text-ren dering-in-opengl-as-of-version-4-1 |
| 09:20 | | RichyB [MyCatVerbs@Nightstar-86656b6c.cable.virginmedia.com] has quit [[NS] Quit: Leaving] |
| 10:44 | | Attilla [Obsolete@Nightstar-3f363270.as43234.net] has joined #code |
| 11:25 | | RichyB [MyCatVerbs@Nightstar-3b2c2db2.bethere.co.uk] has joined #code |
| 11:29 | | Kindamoody is now known as Kindamoody|out |
| 11:31 | | RichyB [MyCatVerbs@Nightstar-3b2c2db2.bethere.co.uk] has quit [Ping timeout: 121 seconds] |
| 12:18 | | himi [fow035@Nightstar-5d05bada.internode.on.net] has joined #code |
| 12:18 | | mode/#code [+o himi] by ChanServ |
| 12:32 | | RichyB [MyCatVerbs@Nightstar-3b2c2db2.bethere.co.uk] has joined #code |
| 12:34 | | Reiver [reiverta@5B433A.3CF6C7.B34AFB.F361B9] has quit [Ping timeout: 121 seconds] |
| 12:35 | | RichyB [MyCatVerbs@Nightstar-3b2c2db2.bethere.co.uk] has quit [[NS] Quit: Leaving] |
| 13:21 | | jeroid [jerith@Nightstar-4426b628.tmodns.net] has joined #code |
| 13:44 | | * TheWatcher grumbles vaguely at DBI::execute()'s handling of extraneous bind parameters. |
| 14:02 | | celticminstrel [celticminst@Nightstar-5d22ab1d.cable.rogers.com] has joined #code |
| 14:16 | <@rms> | At least it's better than PDO's retarded binding issues |
| 14:17 | <@rms> | Hopefully it's better |
| 14:17 | <@rms> | (You can't use the same named bind in PDO twice on the same prepare for some reason) |
| 14:17 | <@rms> | So you end up having to declare variables. |
| 14:17 | <@rms> | (which isn't that bad, but it's still annoying) |
| 14:20 | <@TheWatcher> | Nah, the only issue is that it tries to be too helpful - if you have a prepared query with two bind markers, and you call execute on it with three, it'll fail with an "incorrect number of bind params" error. |
| 14:22 | <@TheWatcher> | Even if the third is actually undef |
| 14:26 | | RichyB [MyCatVerbs@Nightstar-3b2c2db2.bethere.co.uk] has joined #code |
| 14:34 | | You're now known as TheWatcher[afk] |
| 14:40 | < Rhamphoryncus> | Sounds like it's "correct", but still a pita |
| 14:40 | < Rhamphoryncus> | ie you need a way to have a dummy bind marker |
| 15:11 | <@rms> | Heh |
| 15:12 | <@rms> | PDO bitches on that too :/ |
| 15:12 | <@rms> | If it only wans {foo} and {bar} if you give it foo, bar and baz it'll error out |
| 15:13 | <@rms> | PDO is probably the worst lib I've ever used |
| 15:28 | | jeroid [jerith@Nightstar-4426b628.tmodns.net] has quit [Client closed the connection] |
| 15:29 | < Noah> | http://pastebin.starforge.co.uk/519 |
| 15:30 | < Noah> | Is that valid? Assuming that I reload a module that has that line in it, would it import or refresh changes to another module like that? |
| 15:34 | <@rms> | Only way to find out is to try it |
| 15:35 | < Noah> | I suppose so |
| 15:35 | <&jerith> | Noah: You don't really want to use reoal. |
| 15:35 | <&jerith> | Err, reloaf |
| 15:35 | <&jerith> | Ugh. |
| 15:35 | <&jerith> | That thing. Avoid it. |
| 15:35 | < Noah> | Well, the trick is getting the bot to reload certain modules safely |
| 15:36 | <&jerith> | It reloads the thing you reload, but anything referencing the old one stays referencing the old one. |
| 15:36 | < Noah> | And as it stand, it works fine for things in the plugins dir, but any other packages that those plugins import don't get reloaded |
| 15:37 | <&jerith> | Yes. |
| 15:48 | < Noah> | Yea, that didn't seem to work |
| 15:49 | < Noah> | So, how do I go about getting the bot to not only reload plugins, but also any modules that they themselves import...hmm |
| 15:52 | <@rms> | Grr |
| 15:52 | <@rms> | Chrome stops certain security tests like http://127.0.0.1:8125/../../../../etc/fstab |
| 15:53 | <@rms> | (It redirects the user to http://127.0.0.1:8125/etc/fstab) |
| 15:54 | <&jerith> | Run the plugins in a separate process and restart thay. |
| 15:56 | < Noah> | Er...huh? |
| 15:56 | <@rms> | That seems annoyingly expensive |
| 15:56 | <@rms> | Both in programmer-time and computer-time |
| 15:58 | <@rms> | Unless Python has some stupid simple IPC stuff |
| 15:59 | <@rms> | But considering the state of their DB APIs (last I checked), I doubt it. |
| 16:00 | < RichyB> | Python ships with xmlrpclib which is stupidly simple. |
| 16:00 | < RichyB> | Works as an okay RPC if you happen to be targeting certain kinds of frameworks. |
| 16:04 | <@rms> | Yay overhead of parsing XML |
| 16:07 | | * RichyB hunts for the "ignore" button. |
| 16:13 | < Noah> | jerith: How do you mean? |
| 16:15 | < RichyB> | Noah, start a new process, either with the subprocess module or os.fork(), communicate with it over a pipe or socket. |
| 16:16 | < Noah> | Ahhh |
| 16:16 | < Noah> | Okay, I see |
| 16:16 | < RichyB> | subprocess.Popen() may be rather easier if you plan to ever be portable to Windows, I'm not sure how happy the story around the os.fork() function is there. |
| 16:16 | < Noah> | It's on windows currently |
| 16:16 | < RichyB> | Use Popen then. |
| 16:17 | < Noah> | That does complicate a lot of things about my structure though |
| 16:18 | < RichyB> | Ye-es. Alternately, deep reloading isn't impossible, just really fiddly and bug-prone. |
| 16:19 | < RichyB> | Arguably, plugins should usually be confined to separate processes just to stop them taking your main process out if they segfault or crash for some other reason. |
| 16:19 | < RichyB> | Often, but not always, anyway. |
| 16:19 | < Noah> | Well, the context is an IRC bot that's extensible with commands |
| 16:20 | < Noah> | The plugins create the commands and handle logic and make calls to other modules, I could put those modules in the plugin dir too instead as their own package, but I'm trying to keep it sound |
| 16:21 | < Noah> | I could also put the code for the imported packages directly in the plugin scripts, but that's also something I'm trying to avoid |
| 16:21 | < RichyB> | Sensible. I can see why you don't want to complicate things too much. |
| 16:21 | < Noah> | My goal is to make the bot be able to reload all of it's plugin scripts, and the modules that support them |
| 16:22 | < Noah> | That way I can make changes to what the bot does without having to restart it all the time |
| 16:22 | < Noah> | I don't mind having to restart the bot when I change something about HOW it interacts with IRC from twisted, just when I want to make it do something new |
| 16:22 | < RichyB> | Under Unix I might handle this by restarting the whole process without losing the connection to IRC. |
| 16:22 | <@rms> | There is a hacky alternative, it'd require a custom module (in C) though |
| 16:23 | <@rms> | Yeah |
| 16:23 | < RichyB> | (You start a new process, talk to it to hand over the file descriptor corresponding to your IRC connection socket, tell it to go wild on that.) |
| 16:23 | <@rms> | Unless Python supports grabbing an fd from the parent. |
| 16:24 | < RichyB> | I think separate processes is probably your easiest bet. |
| 16:24 | < Noah> | Shoot |
| 16:25 | < RichyB> | Processes aren't even that expensive. New process for every IRC command invocation would be fine for most bots. They're not too complicated unless you want them to be long-running. |
| 16:25 | < RichyB> | Popen([command, line, arguments], stdin=PIPE, stdout=PIPE).communicate("information on stdin")[0] # returns information from stdout |
| 16:26 | | RichyB [MyCatVerbs@Nightstar-3b2c2db2.bethere.co.uk] has quit [[NS] Quit: Leaving] |
| 16:29 | < Noah> | So, basically like that, I would need to structure the modules as if they were command line programs prepared to accept arguements and print out to a screen |
| 16:36 | < Rhamphoryncus> | woot, got my outline shading fixed :D |
| 16:36 | < Rhamphoryncus> | turns out taking the min() of x,y,z is fine for triangles (they represent the distance to each edge), but when trying to also outline the whole square patch you really don't want to include the z.. |
| 16:39 | < Rhamphoryncus> | 2d surfaces don't tend to have useful data in z |
| 16:40 | < Noah> | A side effect of doing it this way is that I can only return text right, no objects |
| 16:42 | < Rhamphoryncus> | easy multiprocess stuff in python? Look at the multiprocessing module :) |
| 16:42 | <@rms> | You can serialize stuff |
| 16:42 | < Rhamphoryncus> | It basically takes some of the threading APIs (like a queue) and layers it over forked processes |
| 16:42 | < Noah> | True |
| 16:44 | < Rhamphoryncus> | The trick to reloading plugins in the conceptually correct way is not *re*loading them, it's *un*loading them |
| 16:44 | < Rhamphoryncus> | If you delete all references to a pure python module it will actually delete it |
| 16:44 | < Noah> | Hmm, good point |
| 16:45 | < Rhamphoryncus> | (C extension modules have some constraints that I don't remember. You don't want to do it.) |
| 16:45 | < Rhamphoryncus> | I've never actually done it but I've always felt you could hook into most of the importing code and basically replace it with your own for your plugins |
| 16:46 | < Rhamphoryncus> | Something where it looks in the plugin dir and either finds a single file (which *cannot* load any other files from that dir) or finds a subdir (which can only load within that subdir) |
| 16:47 | < Noah> | https://bitbucket.org/mao42ranma/mf0dicebot/src/7075a6c5acef/main.py |
| 16:48 | < Noah> | Lines 18 through 33 |
| 16:48 | < Rhamphoryncus> | Any time a plugin needs to store an object externally (such as a callback) you make the API such that you can trivially reset them all |
| 16:49 | < Rhamphoryncus> | And then the best bit: make weakrefs to all the modules that are part of the plugin (one for single file, multiple if a subdir), clear all the references you know of, run the tracing GC, then check the weakrefs to verify they were all cleared. If not you have a module that failed to unload |
| 16:50 | < Rhamphoryncus> | My internet seems broken atm |
| 16:50 | < Rhamphoryncus> | Specifically the router |
| 16:50 | < Rhamphoryncus> | or it's bitbucket |
| 16:51 | < Noah> | blah |
| 16:51 | < Rhamphoryncus> | Anyway that whole scheme is not something you'd accomplish in 5 lines |
| 16:52 | | Kindamoody|out is now known as Kindamoody |
| 16:52 | < Noah> | Try that link again |
| 16:52 | < Rhamphoryncus> | I've played with all the relevant parts of python before so I know it's possible, but that's it |
| 16:52 | < Noah> | I suspect that I had it on private |
| 16:52 | < Rhamphoryncus> | ah works now |
| 16:53 | < Rhamphoryncus> | My DNS is also slow (router issue) and I can't hg push my own tree (ssh connection refused) |
| 16:54 | < Rhamphoryncus> | My own dicebot is a supybot plugin |
| 16:54 | < Rhamphoryncus> | Which uses twisted anyway |
| 16:56 | < Noah> | See, I wonder if I can't creat a list of sorts that contains all the imports that I want to reload, then have the script reloader in main just iterate through that list instead |
| 16:57 | < Noah> | Obviously I don't want it to reload twisted |
| 16:58 | | Netsplit *.net <-> *.split quits: ShellNinja, Attilla, @Derakon[AFK], celticminstrel, Noah, @PinkFreud, froztbyte, @himi, iospacedout, @McMartin, (+7 more, use /NETSPLIT to show all of them) |
| 16:59 | | Netsplit over, joins: @Tamber, @PinkFreud, @ToxicFrog, froztbyte, iospacedout, @himi, &Derakon[AFK], &McMartin, Namegduf, Noah (+7 more) |
| 16:59 | < Noah> | Okay, whatever it's doing currently for the plugins in the plugins dir works, I can change stuff in the script, and the bot uses those changes |
| 17:00 | < Rhamphoryncus> | Yeah, most of the time it will "work", but there's a good chance you leak memory each time |
| 17:00 | < Rhamphoryncus> | Which, honestly, that's probably good enough |
| 17:01 | < Noah> | Well, they aren't 500MB scripts |
| 17:01 | < Noah> | So I think I'd be okay even if they did leak |
| 17:01 | < Rhamphoryncus> | yeah, and you just need to restart it once every 50 reloads |
| 17:02 | < Rhamphoryncus> | The point where things fail is if you're retaining a reference rather than looking things up by name each time |
| 17:03 | < Noah> | I don't suspect that would happen often with what I'm doing |
| 17:04 | < Rhamphoryncus> | yeah |
| 17:04 | < Rhamphoryncus> | it's just a dicebot |
| 17:05 | < Noah> | For now |
| 17:05 | < Noah> | I do plan to do more things with it |
| 17:17 | < Noah> | Well, that doesn't seem to work either. |
| 17:18 | < Noah> | I may just have to settle with keeping all my code in the plugins instead |
| 17:18 | < Noah> | Or at least in the plugin dir |
| 17:19 | | Kindamoody is now known as Kindamoody|out |
| 17:30 | < Rhamphoryncus> | iirc supybot has some sort of unload or reload ability |
| 17:30 | < Rhamphoryncus> | Of course the command & help system is abysmal |
| 17:58 | | Rhamphoryncus [rhamph@Nightstar-5697f7e2.abhsia.telus.net] has quit [Operation timed out] |
| 18:01 | | Vash [Vash@Nightstar-241cb5d4.wlfrct.sbcglobal.net] has joined #code |
| 18:01 | | mode/#code [+o Vash] by ChanServ |
| 18:02 | | Rhamphoryncus [rhamph@Nightstar-5697f7e2.abhsia.telus.net] has joined #code |
| 18:13 | | Attilla [Obsolete@Nightstar-3f363270.as43234.net] has quit [[NS] Quit: ] |
| 18:23 | | You're now known as TheWatcher |
| 18:24 | | Attilla [Obsolete@Nightstar-2f52281f.threembb.co.uk] has joined #code |
| 18:24 | < Rhamphoryncus> | Gah LOL |
| 18:25 | < Rhamphoryncus> | So I'm tweaking some effect options in KDE |
| 18:25 | < Rhamphoryncus> | playing with window wobble, seeing if I can turn it down far enough to be usable |
| 18:25 | < Rhamphoryncus> | Apparently I went to the other extreme |
| 18:25 | < Rhamphoryncus> | It's spazming all over the screen |
| 18:25 | < Rhamphoryncus> | and it WON'T STOP |
| 18:26 | < Rhamphoryncus> | alt-shift-F12 to the rescue |
| 18:27 | < Rhamphoryncus> | and again |
| 18:28 | <@TheWatcher> | 'window wobble', what? |
| 18:28 | <@TheWatcher> | You kids and your fancy gizmos. |
| 18:28 | < Rhamphoryncus> | it's a gimick they dreamed up when they first started playing with opengl window managers |
| 18:29 | < Rhamphoryncus> | You grab the titlebar to drag a window and it lags ever so slightly. It's not rigid anymore |
| 18:29 | < Rhamphoryncus> | Except it's never "ever so slightly". It typically varies between "lots" and "OMFG" |
| 18:30 | <@TheWatcher> | Um.. /why/ |
| 18:30 | < Rhamphoryncus> | because they can apparently |
| 18:30 | <@TheWatcher> | That sounds phenominally pointless |
| 18:31 | < Rhamphoryncus> | Well, originally it was just an experiment |
| 18:31 | < Rhamphoryncus> | "Now that we can do this let's see what works and what doesn't" |
| 18:31 | < Rhamphoryncus> | It's in the "what doesn't" camp |
| 18:34 | < Rhamphoryncus> | There's a lot that can work but it's mostly very subtle |
| 18:35 | < Rhamphoryncus> | Like I went and turned on a slight darkening of non-focused windows. Much easier for me to know which one is the focused one |
| 18:35 | < Rhamphoryncus> | Of course if they just clearly marked that window's title bar like they used to I wouldn't care :/ |
| 18:41 | <~Vornicus> | There are a few window-wobble-ish effects that might be reasonable |
| 18:41 | <~Vornicus> | a slight deformation of corners when you get into a magnet region would work, for instance |
| 18:41 | <@TheWatcher> | Possibly indicating that it needs attention, maybe |
| 18:43 | <~Vornicus> | The key to these is subtlety. |
| 18:49 | | Kindamoody|out is now known as Kindamoody |
| 18:58 | | Kindamoody is now known as Kindamoody[shower] |
| 19:17 | | Kindamoody[shower] is now known as Kindamoody |
| 19:44 | | Kindamoody is now known as Kindamoody[zZz] |
| 19:45 | < gnolam> | http://glyphic.s3.amazonaws.com/ozone/mark/periodic/Periodic%20Table%20of%20the% 20Operators%20A4%20300dpi.jpg |
| 19:52 | < Rhamphoryncus> | That compass rose hurts my brain |
| 20:02 | < froztbyte> | I'm going to just go right ahead and close that tab |
| 20:02 | < froztbyte> | since it has no use in my life :D |
| 20:10 | | eckse [eckse@Nightstar-44362701.dsl.sentex.ca] has joined #code |
| 20:10 | | mode/#code [+o eckse] by ChanServ |
| 20:11 | | Attilla [Obsolete@Nightstar-2f52281f.threembb.co.uk] has quit [[NS] Quit: ] |
| 20:13 | | Attilla [Obsolete@Nightstar-3f363270.as43234.net] has joined #code |
| 21:03 | < Rhamphoryncus> | bwahahahaha |
| 21:03 | < Rhamphoryncus> | KDE's okular (pdf viewer) has an option in the settings for "Obey DRM Restrictions". You can turn it off and suddenly you get thumbnails |
| 21:03 | < Rhamphoryncus> | I bet you can thank adobe for putting that in the spec |
| 21:24 | <~Vornicus> | that's a ridiculous number of operators. |
| 21:35 | < gnolam> | Yes. Yes it is. |
| 21:35 | | * gnolam wonders what TheWatcher has to say about it. |
| 21:50 | | himi [fow035@Nightstar-5d05bada.internode.on.net] has quit [Ping timeout: 121 seconds] |
| 21:52 | | * McMartin likes the prelude |
| 21:54 | <&McMartin> | "Since the Classical Period, knowledge of these operators has significantly increased. They are now spelled differently than they were in ancient times" |
| 21:54 | <&McMartin> | Oh wait, this is Perl 6 |
| 21:55 | <&McMartin> | Does it have any implementations not written in Haskell yet |
| 22:00 | <&McMartin> | Apparently so |
| 22:00 | < gnolam> | Also, I want to see someone set /that/ periodic table to the tune of "I Am the Very Model of a Modern Major-General". |
| 22:18 | <&McMartin> | "Syntactic simplification": "braces now optional, like C" |
| 22:18 | <&McMartin> | That is not what simplification means |
| 22:18 | <&McMartin> | Angry about Perl designers |
| 22:18 | <&McMartin> | on the internet |
| 22:23 | | ToxicFrog [ToxicFrog@ServerAdministrator.Nightstar.Net] has quit [Operation timed out] |
| 22:53 | | ToxicFrog [ToxicFrog@ServerAdministrator.Nightstar.Net] has joined #code |
| 22:53 | | mode/#code [+o ToxicFrog] by ChanServ |
| 23:00 | < Rhamphoryncus> | It's disturbing that they have all of those years of experience with previous perl versions, grafting on more and more, and when they finally decide to make a clean break.. that. |
| 23:01 | <&McMartin> | I still find it hilarious that for years the most conformant implementation was in Haskell |
| 23:01 | < Rhamphoryncus> | heh |
| 23:01 | <&McMartin> | I love Haskell, but when it's leading the way in practical implementations, your problem may be too theoretically complex =P |
| 23:01 | < Rhamphoryncus> | Is.. that a fast forward operator? |
| 23:01 | <&McMartin> | =>>? |
| 23:01 | < Namegduf> | Does it make the program go faster? |
| 23:01 | <&McMartin> | C has >>=. |
| 23:01 | < Rhamphoryncus> | ==>> |
| 23:01 | <&McMartin> | "Shift right and assign back" |
| 23:02 | <&McMartin> | I dunno man |
| 23:02 | <&McMartin> | Also, Perl 6 has been in the design phase for 12 years now |
| 23:02 | < Rhamphoryncus> | Which *is* a fast forward operator. It's a "feed forward" with a seek |
| 23:02 | <&McMartin> | hee |
| 23:02 | < Namegduf> | Hmm. And C++ has >>, the vague directional concept operator for anything vaguely directionally. |
| 23:02 | <&McMartin> | Well Then (tm) |
| 23:03 | <&McMartin> | Namegduf: Remember, kids, they're trained standards committee members, don't try this at home |
| 23:03 | < Rhamphoryncus> | Okay, the only solution here is take away their right to use operators. They can have it back when they start acting like adults. |
| 23:03 | <&McMartin> | The *true* solution is to make operators just be functions with symbolic names that are infix by default, like in ML! |
| 23:03 | <&McMartin> | I'm actually not sure if I'm kidding |
| 23:04 | < Namegduf> | If all language semantics are just function calls, anything can be anything |
| 23:04 | < Namegduf> | Whee |
| 23:04 | <&McMartin> | Though doing certain legal things in ML were also described in the book I learned ML from as leading "only to madness" |
| 23:04 | <&McMartin> | Well, most functions are prefix by default in most languages |
| 23:04 | <&McMartin> | If you have your operators be prefix too, hi Lisp, I didn't see you in that corner where you've been sitting for the past 50 years |
| 23:04 | <&McMartin> | I hear there are some kids on your lawn. |
| 23:06 | <&McMartin> | But yeah, Lisp gives Zero Fucks here and just lets + and - and whatever just be identifiers like any other, bound to sum/difference/etc functions |
| 23:06 | < Namegduf> | Yeah. |
| 23:06 | < Namegduf> | One day I will learn how to write Lisp which doesn't get tangled up in shared state and funky action at a distance. |
| 23:06 | < Namegduf> | Maybe. |
| 23:06 | < Namegduf> | I might just learn Haskell. |
| 23:07 | < Namegduf> | That sounds like the better investment. |
| 23:07 | <&McMartin> | I gotta admit, going back to Scheme post-Haskell was like pulling teeth. |
| 23:07 | <&McMartin> | It really is that much cleaner |
| 23:08 | < Rhamphoryncus> | Hmm. I wonder if we'll have to wait for C++27 before they snarf python3's string .format() scheme to replace the << crap |
| 23:08 | < Rhamphoryncus> | Apparently C++11 snarfs the RNG and has std::mt19937 |
| 23:10 | | maoranma [nbarr@Nightstar-9e797a1c.pools.spcsdns.net] has joined #code |
| 23:11 | <&McMartin> | (Clojure bugged me less, oddly, because Clojure's tightly enoug bound to OO-like VMs that I don't keep reaching for Haskell control or data constructs in the first place) |
| 23:12 | | Noah [nbarr@Nightstar-9e797a1c.pools.spcsdns.net] has quit [Ping timeout: 121 seconds] |
| 23:16 | | maoranma is now known as noah |
| 23:16 | | noah is now known as Noah |
| 23:18 | < Rhamphoryncus> | Haw. The standard mandates that the MT RNG produce certain values at certain times XD |
| 23:19 | < Rhamphoryncus> | (not as silly as it sounds, since it's a PRNG and that's with a specified seed value) |
| 23:21 | < celticminstrel> | Um. How does "braces now optional" get followed by "like C"? |
| 23:22 | <&McMartin> | It's optional for single-line clauses. |
| 23:22 | <&McMartin> | Perl did not permit while (i < 10) ++i; before, you needed while ($i < 10) { $i += 1; } |
| 23:26 | | ShellNinja [abudhabi@Nightstar-7b2d96ca.adsl.inetia.pl] has quit [Ping timeout: 121 seconds] |
| 23:27 | | cpux [cpux@Nightstar-c5874a39.dyn.optonline.net] has joined #code |
| 23:28 | | * Rhamphoryncus peeks at boost::format |
| 23:28 | < Rhamphoryncus> | makes me want to reimplement python's formatting |
| 23:31 | < Rhamphoryncus> | with pformat() and errformat() helper functions for stdout and stderr, since those are excruciatingly common usage |
| 23:32 | | himi [fow035@D741F1.243F35.CADC30.81D435] has joined #code |
| 23:32 | | mode/#code [+o himi] by ChanServ |
| 23:48 | < Rhamphoryncus> | Oh look, an example of C++11 variadic templates that actually makes sense: http://oopscenities.net/2011/07/19/c0x-variadic-templates-functions/ |
| 23:49 | < Rhamphoryncus> | So it is in fact possible to implement a sane printf()-equivalent in C++11 |
| 23:51 | | * gnolam reposts http://alan.dipert.org/post/153430634/the-march-of-progress |
| 23:54 | < Rhamphoryncus> | yeah |
| 23:55 | < Rhamphoryncus> | There's significant drawbacks with C's scheme, but if you want to replace it with something better you have to actually *replace it with something better* |
| 23:57 | < Rhamphoryncus> | In classical C++ you can't do a type-safe printf(). There's no type-safe variadic scheme, so you'd have to have 32 overloads, one for each length of argument list, and just not allow more than that. Oh wait, that wouldn't be bad at all :P |
| --- Log closed Tue May 22 00:00:53 2012 |