00:00:29 Sgeo: great 00:01:37 elliott@dinky:~/code/pkg$ wc -l pkginfo.c 00:01:38 174 pkginfo.c 00:01:38 elliott@dinky:~/code/pkg$ wc -l pkginfo.sh 00:01:38 39 pkginfo.sh 00:01:46 pikhq: I think that maybe an sh-based packaging utilities system is a splendid idea. 00:03:19 <3 this game 00:03:30 Played the single-player campaign before, but still fun 00:07:49 elliott: Mmm. 00:08:53 pikhq: Even though sh is terrible. 00:13:13 pikhq: 00:13:13 $ busybox sh pkginfo levee | awk '/^Version:/ { print $2 }' 00:13:13 3.5a 00:13:17 I declare it portable and useful. 00:22:20 pikhq: Virtual packages: http://sprunge.us/SdDH 00:25:02 pikhq: I should probably write the actual installation script in C so that if you replace the utilities a shell version would use it still works... 00:29:37 -!- FireFly has quit (Quit: swatted to death). 00:39:44 pikhq: I use $*. Question: Am I bad? 00:40:15 elliott: you're talkative 00:40:19 elliott: http://www.jaybirdgear.com/cart/sb2/# 00:40:21 which color 00:40:53 augur: they look uncomfortable, putting form above function, and lookinga t it they probably have terrible audio quality. 00:40:56 oh bluetooth? 00:41:03 it compresses the audio lossily then 00:41:09 have fun with that 00:41:53 :| 00:45:07 Why does Perl have an unless? 00:45:37 Sgeo: Because it does. 00:46:17 Sgeo: because it reads better 00:46:19 same as "until" 00:48:10 elliott: $*? Not necessarily bad. 00:48:25 naked blocks should not count as loops 00:48:33 Sgeo: Good! Because they don't. 00:48:33 They should count as ... scopy things 00:48:38 Sgeo: Good! Because they do. 00:48:52 elliott: And, yeah, having the installer in C is a good idea. 00:48:52 The slide for my class says: 00:48:57 Provides an "oh shit" button. 00:49:02 There are five kinds of loop blocks in Perl. These are the blocks of 00:49:02 for, 00:49:02 foreach, 00:49:02 while, 00:49:02 until, or 00:49:03 the naked block. 00:49:13 pikhq: A good idea that I am ignoring because it's too much of a pain! 00:49:15 pikhq: ~patches welcome~ 00:50:08 elliott: At bare minimum, provide an easy way to run it using busybox. 00:50:19 And make it function using the busybox you ship, of course. 00:50:37 pikhq: Why? What's more reliable about busybox vs. ksh? 00:51:01 elliott: Getting a full busybox running is easier than a full coreutils. 00:51:33 Well, the loop control... stuff (like redo) works in naked blocks 00:51:34 In case of dire emergency, I could readily statically link a busybox from just about any system. 00:51:51 pikhq: http://www.wormhole.hu/~ice/ksh/ 00:51:55 Of course, you *could* just say "screw it", like Gentoo does. 00:51:59 @d unless(_1) if(!(_1)) 00:52:02 @f unless if 00:52:03 pikhq: Find me a system with a C compiler and BSD make that this doesn't compile on. 00:52:05 (which requires a full Python installation for the package manager) 00:52:06 pikhq: (Hint: Such a system does not exist.) 00:52:18 elliott: Hint: your package manager relies on more than *just* a shell. 00:52:36 pikhq: Hint: it's rather likely that the rest of the coreutils is busybox. :p 00:52:40 In my opinion the package manager should not require Python 00:52:49 elliott: Okay, so make it function perfectly with ksh and busybox. 00:53:14 elliott: That is still *quite* easy to set up. 00:53:30 pikhq: "busybox sh" is my Portability Checker(TM). 00:53:32 ... Heck, if your package manager relies on just those two, you could probably bootstrap a Kitten system from a floppy disk. 00:53:50 Ah. Yeah, if it works on Busybox sh, it'll probably work anywhere. 00:53:55 pikhq: (Also dash, because it's so terrible that nothing that runs on it could possibly fail anywhere else.) 00:53:56 What with being a *subset* of Bourne shell. 00:54:40 Mmm, having a minimal installer smaller than Debian. :P 00:55:34 * Sgeo wants @ 00:55:51 I'm pretty certain that this discussion is not about @ 00:56:04 Sgeo: No, it isn't, but no way am I gonna develop @ on Debian. 00:56:10 I want something I can trust not to go stupid overnight. 00:57:00 pikhq: Hmm, does using rsync to update a compressed tarball work well? 00:57:07 pikhq: Obviously updating a tarball with it is just as good as updating each individual file. 00:57:19 pikhq: But do, say, .tar.xzs change more than the set of changes to the original files? 00:57:29 Thus making rsync less efficient on these already-compressed things? 00:57:39 elliott: It would not work well with a compressed tarball. 00:58:12 pikhq: Are you suuuuure? 00:58:16 For the *precise* same reasons it doesn't work well with encryption. 00:58:35 pikhq: Right, because perfectly compressed data is indistinguishable from random data. 00:58:51 But that doesn't mean you can't diff two pieces of similar random data and get a short diff, no? 00:59:03 They won't be similar, in general. 00:59:20 pikhq: Well, I'm also not going to run unxz before every rsync and xz afterwards. 00:59:26 pikhq: Suggestions welcome. :p 01:00:37 Store uncompressed, use rsync's compression? :P 01:01:01 pikhq: Store uncompressed = lol, wasteful. I guess it isn't a big deal, but still. 01:01:18 pikhq: (Also, fun fact: I'm compressing the kernel with LZO for the sole reason that it's the fastest to decompress.) 01:03:01 elliott: http://zsync.moria.org.uk/ 01:04:20 pikhq: Nice! gzip-specific... but better than nothing. 01:04:33 elliott: Actually, the gzip-thing could be readily patched. 01:04:41 pikhq: Yeah, but I have enough work on my hands anyway. 01:04:47 Mmkay. 01:05:01 [[If you don't specify the names of any files to put in the archive, then tar will create an empty archive. So, the following command will create an archive with nothing in it: 01:05:01 tar --create --file=empty-archive.tar]] 01:05:02 LIES 01:05:05 tar: Cowardly refusing to create an empty archive 01:05:29 But hey, it's rsync-like and works on compressed files, and seems to work much nicer with standard (easily available) servers. 01:06:06 ha, apparently you do 01:06:09 --files-from=/dev/null 01:06:12 to stop tar trying to second-guess you 01:06:22 what an anti-feature 01:06:39 wtf, that just creates a bunch of zeroes 01:06:43 10K of them 01:07:02 well, whatever 01:09:31 pikhq: Things I hate more than anything: Stuff that remembers its --prefix. 01:09:40 I WANT TO BE ABLE TO COMPILE SHIT WITHOUT USING A CHROOT, FUCKERS 01:11:12 pikhq: Holy shit, you know fakeroot? 01:12:08 elliott: Yeah? 01:12:28 pikhq: Guess what version control system it uses. 01:12:29 pikhq: Wrong! 01:12:35 ... 01:12:36 pikhq: The answer is Arch 1, aka tla. 01:12:37 Patch? 01:12:42 *Oh*. 01:12:44 *vomit* 01:12:45 flrghl 01:12:45 pikhq: *The thing uses Arch. In 2010.* 01:12:50 *arch 01:12:58 I'm sorry, but, I have to respect anyone crazy enough to do that. 01:13:05 % tla register-archive fakeroot@packages.debian.org--fakeroot \ 01:13:05 http://arch.debian.org/arch/fakeroot/fakeroot/ 01:13:05 % tla get -A fakeroot@packages.debian.org--fakeroot fakeroot--main--0.0 01:13:23 tla's motto is "Simple things should be difficult, hard things should cause you to commit suicide." 01:13:28 *difficult things 01:15:34 extern.h:200: error: conflicting types for ‘getline’ 01:15:34 /usr/include/stdio.h:651: note: previous declaration of ‘getline’ was here 01:15:42 pikhq: in which glibc extensions harm someone not even using them 01:16:16 elliott: -std=c99 should turn off glibc extensions. 01:16:24 pikhq: *c89 01:16:25 in this case 01:16:26 Indeed. Still. 01:16:42 Still, the extensions should *not* be on by default. 01:16:56 Have them all you want, but they should be enablable options, not default. 01:17:10 there should be a #define to disable them 01:17:15 or an #undef 01:17:29 GreaseMonkey: There's a #define to enable them. That GCC does by default. 01:17:36 You may now curse at GNU. 01:17:53 Maybe the GNU extensions should be disabled if the compiler is invoked by the name 'cc' instead of 'gcc' 01:18:39 pikhq: Wait a second, why the fuck does the C compiler change what the glibc does by setting how the code is compiled? Does glibc actually check if the C compiler claims to implement "GNU C", and help it in its evil quest by adding features of GNU C? 01:18:43 hrfffffffffff gnu lockin 01:19:20 I think standards mode controls predefined macros and macros control glibc evilness 01:20:14 elliott: Glibc checks if the C compiler claims to implement "GNU C", and helps it in its evil quest by adding featurs of GNU C. 01:20:29 elliott: BTW, another option: gzip --rsyncable. 01:20:42 elliott: Glibc checks if the C compiler claims to implement "GNU C", and helps it in its evil quest by adding featurs of GNU C. 01:20:45 ha ha ha ha ha ha ha i hate this shit. 01:20:48 elliott: Produces gzip output that is *barely* larger, but readily rsyncable. 01:21:05 pikhq: Nice. But I think I'll just use zsync anyway. :p 01:21:09 HTTP access is nice, after all. 01:21:28 Yeah; quite a bit nicer to do it all with nothing more than a somewhat-sane HTTP server. 01:25:26 $ sudo ./pkginstall levee 01:25:26 * Installing foo... 01:25:26 * Installing levee... 01:25:28 HAHAHA SCIENCE 01:26:26 pikhq: Dependencies of that: the "set" built-in, the "[" built-in, the "cd" built in, echo, tar, and touch. 01:28:12 Beautiful. 01:28:16 -!- zzo38 has quit (Quit: This is a secret message! Please figure out what it means!). 01:28:45 elliott: How's it doing dependency resolution? 01:28:50 Naively? 01:28:56 if [ -e $1/needs ]; then 01:28:56 for dep in $1/needs/*; do 01:28:56 install_pkg $(readlink $dep) 01:28:56 done 01:28:56 fi 01:28:59 pikhq: Is that naive enough for you? 01:29:03 Naively. 01:29:07 pikhq: Of course install_pkg does nothing if it's already installed. 01:29:20 pikhq: Feel free to suggest a better way, but I'm unlikely to use it because... there are no runtime dependencies. 01:29:27 Termcap files. That'd be a runtime dependency. Apart from that, uh... 01:29:29 Though likely working just fine for your purposes, because you are unlikely to ever have circular dependencies. 01:29:34 Executables that other programs use, obviously. 01:29:40 But ... not much really. 01:30:02 Yeah, the odds are that your dependency graph won't get ridiculously complex. 01:30:31 And that anyone who really wants "standard" programs can install Gentoo Prefix or something. 01:31:44 pikhq: Define "standard" programs. 01:33:03 Expecting your standard, dynamically linked, GNU-filled environment. 01:33:22 Ew. 01:33:41 pikhq: "But why would you want to". :p 01:34:03 pikhq: Although I'm hardly unbiased; I'm including Java for one single program. Guess what program. 01:34:13 Minecraft. 01:34:23 pikhq: HOW DID YOU GUESS 01:34:25 :p 01:34:36 MAGIC 01:34:46 elliott: Java? :/ 01:34:55 olsner: Minecraft is written in Java. 01:35:05 I will not run any OS that cannot run Minecraft. 01:35:06 Thus Java! 01:35:37 Back. 01:35:39 well... wtf? 01:35:45 olsner: What do you mean, "wtf?"? 01:36:07 elliott running any kind of java thing, if it weren't for minecraft it would be inconceivable 01:36:27 olsner: i have a slight feeling you're trolling me here :) 01:36:32 olsner: I KNOW YOU'VE USED ANT 01:37:23 hmm, no, I'm not really trolling 01:38:34 I have never used ant voluntarily, note you 01:39:05 but I have indeed used it, which is how I know how horrible and braindead it is 01:39:46 it's kind of bancstar+xml+java molded into a build system 01:40:29 Ant is perhaps the worst-designed build system I have seen. 01:40:52 ... "build system" - at best you can use it to do sequencing in a language worse than dos/cmd batch files 01:40:55 you've never used drupal then 01:41:17 And I include the horrid monstrosity I once saw that was a hybrid of *interactive* Perl, shell, Make, Autotools, and C in that consideration. 01:41:32 pikhq: git? 01:41:35 cheater99: No. 01:41:38 cheater99: IRAF. 01:41:49 what you have just described sounds like git. 01:41:50 cheater99: Which is also notable for being almost its own OS. 01:41:58 even a "do your worst" given all those tools should end up better than ant 01:42:11 -!- Mathnerd314 has joined. 01:42:11 ==Phantom_Hoover logreading== I have struck diamond. I WILL NOW ACCOUNT FOR EVERY SINGLE DIAMOND BLOCK I MINE SO THAT IT CAN BE SPLIT 50/50. 01:42:14 olsner: Quite true; Ant is just revolting. 01:42:18 Even though it matters not. 01:42:29 2 blocks. 01:42:30 pikhq: I am glad we agree :D 01:42:37 + 2 blocks. 01:42:46 + 2 blocks. 01:42:53 + 2 blocks. 01:43:11 Done; 8 blocks in total. 4 each. I will put them in my chest. 01:43:40 http://en.wikipedia.org/wiki/Apache_Ant The sample build file here makes me want to vomit. 01:44:11 It actually demonstrates just about everything wrong with XML. 01:44:36 pikhq: lol - it gets MUCH WORSE 01:44:48 that's not even a real-world example 01:45:01 It has such a revolting amount of syntactic noise. 01:45:19 tag based languages are in general a mega failure 01:45:24 hadn't found one that doesn't suck 01:45:37 Even just replacing the structure with sexps naïvely would dramatically improve that. 01:46:03 with what now? 01:46:05 8 iron blocks mined. 01:46:18 cheater99: With sexps. Lisp syntax. 01:46:26 ant is sufficiently braindead and horrible without the syntactic overhead though 01:46:32 don't think that's the only fault! 01:46:33 ok 01:46:42 olsner: True, I'm *only* criticising the syntax ATM. 01:47:03 it's just... vomit all the way down 01:47:05 olsner: let's write a multithreaded pi computation in ant 01:47:11 olsner: just to prove how bad it can get 01:47:23 Tag-based markup languages are truly vomitous for any usage other than actual markup of text. 01:47:44 cheater99: no. ant is not even interestingly horrible 01:47:57 ok 01:48:04 6 iron blocks. 01:48:27 -!- Sasha has quit (Remote host closed the connection). 01:48:28 i'm listening to music which was sequenced from analysis data of a dna protein 01:48:37 it's actually pretty cool. 01:48:53 -!- Sasha has joined. 01:49:02 cheater99: it's up to you to go ahead and do it, but only do it if you truly have *nothing* better to do 01:49:03 2 iron blocks. 01:49:21 olsner: i have a non-empty set of things better to do 01:49:36 cheater99: I'm glad :) 01:49:39 olsner: Why is it that every make replacement has *worse* syntax? 01:50:04 pikhq: because make is actually at the upper end of build system sensibility 01:50:12 Sadly, it is. 01:50:13 A bunch of coal mined, but whatever. 01:50:29 lots of people seem to say "make sucks", then go build something so much worse 01:50:32 olsner: that is more a statement about everything except make, than a statement about make 01:50:43 elliott: right you are 01:50:49 olsner: well, make works only on some systems 01:51:00 olsner: ant works on more systems 01:51:14 cheater99: that's just the anal sex argument 01:51:20 People should look at what make does *right* when replacing it. 01:51:25 [Sever] COBOL IS A HORRIBLE LANGUAGE. 01:51:28 olsner: no 01:51:29 *Server 01:51:32 Cobol is horrible, Netcraft confirms it. 01:51:42 olsner: people needed ant for things which didn't exist on, say, windows at that point 01:51:43 or on mac 01:51:44 olsner: I like how it's now transformed into "the anal sex argument". 01:51:50 Oh yeah, you know, that common argument about anal sex. 01:51:51 For instance: its syntax is *simple and readable*, and it is *declarative*. 01:52:26 elliott: it isn't afaik, I just thought the analogy was so right for this case 01:52:32 olsner: it is as of your message 01:52:58 cheater99: The build system that shipped with the official Mac development stuff was a Make derivative. :) 01:53:22 cheater99: And Windows has had Microsoft nmake for ages. 01:53:41 pikhq: not before osx 01:54:23 cheater99: are you saying that software was not built on windows before ant? 01:54:31 cheater99: Macintosh Programmer's Workshop. Make and Shell. 01:54:39 pikhq: troll, feeding, etc. 01:54:42 it was, but people needed something interoperable 01:55:00 make vs. shell script, who wins? 01:55:08 elliott: boring, monotnous, etc. 01:55:14 note initial expressing of ant ignorance and then switching to advocating ant when that seems to be the thing that'll generate more messages. 01:55:15 +o. 01:55:17 And anyways, *nowadays* it is very very easy to make an interoperable build system. 01:55:19 Assume UNIX. 01:55:39 elliott: ant ignorance? i use ant in my job. 01:55:43 you know, that thing you don't have. 01:56:57 doubly maintaining separate build systems always existed. makefile/project/whatever generators are also available. also, cygwin (for shell scripts while still supporting windows) and scripting languages 01:57:19 in other words: there is no problem that any solved that didn't already have a better solution 01:57:21 cygwin sucked very much when ant was being created 01:57:25 any/ant 01:57:27 still sucks much 01:57:52 no part of cygwin has suckage in the amounts that ant does 01:57:57 I'm fairly certain that just having two completely different build systems sucks less than ant. 01:57:58 yeah 01:58:00 i agree 01:58:10 i'm just reminding you of the premise :p 01:58:10 (make and Visual Studio) 01:58:32 (or just saying "screw it" and build with mingw) 01:58:38 haha 01:58:40 * GreaseMonkey <3 mingw 01:59:24 cheater99: am i meant to feel inferior because i do not yet have to work to sustain myself? 02:00:31 only if you want to 02:05:40 -!- Hiant has joined. 02:05:49 good night, and beware of ant 02:06:43 nn 02:08:06 I'm fairly certain that just having two completely different build systems sucks less than ant. // build WebKit some time. 02:08:14 It has at least four separate build systems. 02:08:53 More coal, blah blah blah. 02:12:28 6 irons. 02:13:00 More coal. 2 irons. 02:16:12 1 gold; coal. 02:19:47 ok, gotta clone this hdd. bbl sweeties 02:19:58 -!- cheater99 has quit (Quit: Leaving). 02:24:16 Gregor: As in, one build system for each OS. 02:24:25 More. 02:24:26 Gregor: Not an unholy hybrid of build systems. 02:24:34 And an unholy hybrid. 02:24:39 Gregor: He means that one build system is what he prefers to ant. 02:24:43 Thus invalidating your example. 02:24:45 * elliott interpreter 02:25:03 Lemme guess. A large part of it is KDE-autotools. 02:25:18 pikhq: On Mac OS X, each subsystem is independently maintained in a separate XCode build, but only JavaScriptCore will build in XCode. They have to be cobbled together by a series of convoluted Perl scripts. 02:25:23 -!- Hiant has quit (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539]). 02:25:28 (KDE's usage of autotools is sufficiently unique to count as a seperate build system, IMO) 02:25:37 pikhq: It uses autotools for GTK+, qmake for Qt. 02:25:53 ... So they redid the entire KHTML build and made it worse. 02:25:57 Gag me. 02:25:59 3 diamond. 02:26:06 pikhq: KDE uses cmake. 02:26:23 elliott: KDE 3, which Webkit was forked from, used KDE-autotools. 02:26:26 pikhq: Something else for ETL, it has two build systems for Windows (one for VS, one for MingW/Cygwin), and a wholly separate build system based on scons for Chrome. Also it has cmake stuff in there I haven't been able to identify the user of. 02:26:34 pikhq: Right. Although I think late KDE 3 used cmake. 02:26:40 Which had a gigantic swath of its own macros to replace most of KDE 3's functionality. 02:26:42 elliott: No. 02:26:53 And I think the wx fork must have went away since I haven't seen it lately. 02:26:59 elliott: The Cmake switch came into being well after porting to Qt 4. 02:27:28 Erm, s/KDE 3's/autotools'/ 02:34:17 How to become suicidal: 02:34:25 1. Mine endlessly in Minecraft. Pick up tons of valuable materials. 02:34:31 2. Discover a HUGE, and ore-rich, cavern system 02:34:34 3. Get lost. 02:34:41 4. Find what you think is a path you haven't tried yet. 02:34:44 5. There's lava on the way. 02:34:56 6. Successfully execute a jump onto the platform in the middle! Fuck yeah! 02:35:02 7. Jump forwards to get out. 02:35:04 8. End up in the lava. 02:35:08 9. Flail wildly. 02:35:10 10. Burn. 02:35:11 11. Die. 02:35:14 12. Lose everything. 02:40:42 "Lose everything"? 02:40:52 Sgeo: Die -> lose inventory. 02:40:57 :/ 02:41:04 Including the tons of stuff I mined, all my trinkets, and all my tools. 02:41:11 I have enough to rebuild them with some tree-punching in my chest, but still, fuck that shit. 02:41:22 I should have just poured water over it. ...Why didn't I...? EURGH YASD 02:42:14 Holy fuck. EFL uses cmake. But it's not the ONLY user of cmake. I don't know what the other one is. 02:42:19 This has at LEAST a dozen build systems. 02:42:55 Gregor: You are so making me not want to package anything that uses WebKit :P 02:43:04 elliott: Yeah. Don't. 02:43:07 Gregor: I bet it's two times the fun if you want to create static libraries, right? 02:43:18 999999999999999M libwebkit.a 02:43:24 I doubt very highly that it's capable of that. 02:43:36 Gregor: I am pretty sure it is, actually. 02:43:38 Although the GTK+ build definitely does create them since it uses .la 02:43:41 At least the suckless guys seem to think so. 02:43:44 But I doubt it's capable /in general/ of doing that. 02:43:56 That is: Only 1/3rd of its hundreds of build systems support that ;) 02:44:00 Gregor: Thankfully, I'm doubly uninterested in building non-gtk versions :P 02:44:14 And most make/cmake stuff will let me just pass -static and go on with my life. 02:44:23 *get on 02:44:32 Oh look, Android uses a different build system from Chrome, and is Make-based. 02:44:48 Y'know what's terrible? I am currently NOT looking for build systems. I'm trying to do something else. 02:44:57 http://terryxu2008.blogspot.com/2008/08/webkit-runs-on-fltk.html 02:45:10 "Hey guys! I ported WebKit to FLTK, which is a major and useful development! Now watch me never say anything about it again, or release the source!" 02:45:19 Just a WIIIILD weekend hack! 02:45:53 Good, don't need yet another build system :P 02:47:09 http://c2.com/cgi/wiki?WhyNobodyDeletesWiki -- voted for title most likely to cause harm to the wiki :P 02:47:22 WhyNobodyPressesEditAndSelectsAllAndThenPressesSaveOnThisPage 02:47:26 erm 02:47:30 WhyNobodyPressesEditAndSelectsAllAndDeletesAndThenPressesSaveOnThisPage 03:05:24 Vorpal: this synthesis thesis is *wonderful*! 03:05:27 thank you for pointing me to it 03:05:29 i will read all of it 03:10:37 LightBot 2.0 has a level editor 03:11:11 And recursion and conditionals, don't remember if the first had either 03:11:28 * Sgeo goes to play a bit of RoboZZle 03:12:27 Meh 03:12:33 My brain seems to be off right now 03:14:13 How would I clear a screen in brainfuck... 03:15:37 Goosey: You can't. But if you assume a standard terminal -- 25 lines on Windows, 24 on Unix -- then you can just print that number of newlines to be sure you're on a blank screen. 03:15:43 Of course you might be further down than that. 03:15:49 There is ANSI codes and stuff you can output but bleh. 03:16:13 Goosey: If you keep track of how many multiples of 80 chars you have printed (because it wraps the line at 80 chars), and how many newlines, you can do (24 or 25) - that to get how many lines you need to output. But is it really worth the bother? 03:16:18 elliott, how do commands like clear and cls work? 03:16:24 the same way? 03:16:29 Goosey: cls, I don't know. 03:16:36 $ clear | cat -v; echo 03:16:36 ^[[H^[[2J 03:16:40 Goosey: clear, as you can see, uses special control code. 03:16:44 there, ^[ = the escape character 03:16:45 so that's 03:16:50 {ESCAPE}[H{ESCAPE}2J 03:16:59 will clear an ANSI-compliant screen 03:17:03 ah 03:17:14 That makes sense 03:17:24 Never thought I'd type man can 03:17:26 Goosey: escape is ascii 27 03:17:30 man cat for that matter 03:17:55 Goosey: so basically print 27, print "[H", print 27, print "2J". you probably want to print 10 after that for new-line so that it gets flushed to the terminal; otherwise it might be buffered and not happen 03:18:06 hmm 03:18:08