00:18:34 -!- FireFly has quit ("Later").
00:20:49 -!- augur has joined.
00:32:08 -!- oerjan has quit ("leaving").
01:36:05 -!- Asztal has quit (Read error: 110 (Connection timed out)).
02:21:18 -!- fax has quit ("Leaving").
03:14:11 -!- zzo38 has joined.
03:14:44 <zzo38> In order to learn to be a Real Programmer, one should write a Forth interpreter in machine-codes (or assembly language).
03:16:11 <zzo38> One day, in order to learn how to build a computer, I will try to make a computer with z80-based, and a On-Screen-Display chip, a few buttons on the front, keyboard connect, audio, and external memory.
03:16:17 <pikhq> In order to be a Real Programmer, one should learn a defining feature of programmers: laziness.
03:16:58 <zzo38> I can have addresses 0000-3FFF=internal ROM,4000-7FFF=RAM, 8000-BFFF=external memory,C000-FFFF=I/O and OSD
03:17:44 <zzo38> In order to do it, I should write a emulator, with a option to enable/disable ROM writing, to make the ROM act as RAM instead if you enabled that option, so that it can be written a Forth interpreter with some stuff built-in
03:18:17 <zzo38> Much of Forth is generally written in Forth itself, so therefore a emulator with that option would be a useful feature to make a Forth system from it.
03:19:06 <zzo38> pikhq: Anyone who should be a Real Programmer should learn a few things about machine-codes!
03:19:26 <pikhq> zzo38: I'm not disputing that.
04:39:14 <Warrigal> TI graphing calculators use z80-type processors, don't they?
04:53:55 <Sgeo> Hm, in this spec I'm writing, I don't define what I mean by an "internal key"
04:54:29 <Sgeo> I mean, it hardly makes sense to define it in this spec, since the way my product uses it will require more than just this spec
04:54:53 <Sgeo> But not defining it would lead to compatibility issues with anyone who wanted to compete and read my data in
05:11:52 <zzo38> What is this spec for?
05:12:08 <Sgeo> zzo38, an upcoming product of mine.
05:12:31 <Sgeo> zzo38, http://docs.google.com/View?id=dgtq868h_62gp6gxtz3
05:12:41 <Sgeo> I'm going to stop publishing soon, so read it
05:13:32 <Sgeo> "An animation list consists of animation names, followed by spatial specification, possibly followed by camera specification, separated by colons, and an amount of time, in integer milliseconds" needs to be reworded
05:13:54 <Sgeo> And the EBNF is incomplete, but afair accurate
05:15:08 <zzo38> I'm still unsure what its purpose is
05:16:38 <Sgeo> The product will allow people to set up positions where people who sit on furniture end up sitting
05:16:51 <Sgeo> Each "seat" can play several animations in order
05:16:58 <Sgeo> With camera changes
05:18:23 <Sgeo> It just occured to me that the spec version shouldn't be specified per seat
05:19:27 <Sgeo> But still accessible WTF?
05:20:03 <zzo38> Are control characters allowed in animation names?
05:20:45 <Sgeo> You mean ASCII control characters? No, because SL doesn't allow it
05:21:17 <zzo38> So it need not be specified in this spec
05:23:55 * Sgeo still has stuff to add to the spec
05:28:28 <Sgeo> Ok, going to eat now
05:28:31 -!- mad has joined.
05:28:51 <mad> anyone into conlangs?
05:33:06 <mad> If you could add a sound or distinction into English, what would you add? :D
05:47:51 <zzo38> My ideas of conlangs is a bit of things. One, is that to represent sounds, you would have shapes for "glottal" and "plosive" and "fricative" and so on, and for a glottal stop you somehow combine the shapes for "glottal" and "plosive" together
05:49:11 <zzo38> And then you can add a voice mark (similar to how Japanese has a voice-mark for a voice-sound)
05:49:52 <mad> what about t+glottal :D
05:50:13 <zzo38> I don't know about that
05:51:54 <mad> korean has something kinda like what you're saying
05:52:16 <mad> th (aspirate t) is like t with an extra bar
05:52:42 <mad> and ph, chh, kh (all aspirate too) have the same extra bar
05:54:57 <zzo38> Yes, but not what I mean. The "voice mark" in Japanese converts "t" to "d"
05:55:47 <mad> well, it's the same thing, except the voice mark is less part of the letter
05:57:37 <mad> some transcription schemes use 'h' for "no voicing" for letters that don't normally have an voiceless version
05:57:53 <zzo38> But I don't know Korean so I wouldn't always know
06:02:18 <mad> hmong has hm, hml, hn, hny, hl for instance
06:03:23 <zzo38> Except, Japanese also has a semi-voice-mark, converting 'h' to 'p' (voice-mark converts 'h' to 'b'). Japanese language also uses small 'tsu' to represent a pause (which acts just like any other letter, it can be part of a word)
06:03:59 <zzo38> Both these things might seem strange at first, until you look at many Japanese writing, and you can see how the words are constructed, then it appears logical.
06:04:22 <mad> yeah, that's due to how japanese evolved
06:04:48 <mad> obviously a semi-voice mark could not apply to english
06:05:14 <zzo38> Basically same thing as I mentioned. By reading many Japanese writings, I can see how the logic works why it is like this.
06:05:42 <mad> yeah, usually writing systems follow how languages work
06:07:26 <zzo38> (The Japanese writing I read mostly is the Akagi manga books)
06:09:31 <mad> how much japanese do you know%?
06:09:57 <mad> do you think english should be more like japanese? :D
06:10:08 <zzo38> Not a lot, but more than many people do
06:10:22 <zzo38> I have no opinion on whether or not English should be more like Japanese.
06:10:45 <zzo38> English already exists the way it is, and probably shouldn't be change too much except for slight evolve
06:11:04 <zzo38> If you want differently, you can make conlangs (which was what you first asked about, anyways)
06:11:29 <mad> what, some people think english should be more like japanese?
06:11:45 <zzo38> Actually I don't know.
06:12:08 <zzo38> I don't know whether or not some people think English should be more like Japanese.
06:12:36 <mad> japanese has a nice feature or two (pronunciation, simple noun morphology) but it more than makes up for that in other ridiculously complex parts (verbs, writing, etc)
06:13:07 <zzo38> Well, yes... of course it is. Any language has some wrong parts!
06:15:23 <mad> yeah but some have more :D
06:19:31 <mad> some have huge pronunciation pitfalls or head spinning grammar
06:26:20 -!- zzo38 has quit (Remote closed the connection).
06:58:07 <pikhq> Verbs, complex in Japanese?
06:59:21 <pikhq> ... Says "natteirareta".
07:00:38 -!- zzo38 has joined.
07:00:51 <zzo38> asiekierka asked about game of life algorithm
07:01:58 <zzo38> One algorithm (I don't know whether it meets your criteria or not), is just a simple one: a convolution filter followed by a CLUT
07:02:56 <zzo38> ImageMagick will do both.
07:07:47 <mad> pikhq: well, more like all the system of implicit subjects and objects and stuff :D
07:17:56 -!- zzo38 has left (?).
07:26:40 -!- Sgeo has quit (Read error: 110 (Connection timed out)).
07:59:59 -!- clog has quit (ended).
08:00:00 -!- clog has joined.
08:25:59 -!- mad has quit ("Radiateur").
09:33:36 -!- oerjan has joined.
09:54:49 -!- FireFly has joined.
10:09:00 <fizzie> Hrm, my embedded rfk86 font (after generating a SVG vector font out of the bitmaps, and converting that to truetype with fontforge) didn't turn up so great: http://zem.fi/~fis/ugly.png
10:10:02 <fizzie> The SVG font metrics should make it very very monospaced, but maybe fontforge's doing something silly to it while converting.
10:26:31 -!- Pthing has joined.
10:38:25 <FireFly> well, SVG fonts works as webfonts, don't they?
10:42:53 -!- oerjan has quit ("leaving").
10:43:13 -!- kar8nga has joined.
10:53:58 <fizzie> Firefox didn't even render a SVG with embedded SVG font.
11:18:56 -!- kar8nga has quit (Remote closed the connection).
12:17:39 -!- clog has joined.
12:17:39 -!- clog has joined.
11:54:27 -!- clog has joined.
11:54:27 -!- clog has joined.
14:05:01 <Deewiant> The tubes are once again clogged
14:06:29 -!- fax has quit ("Leaving").
14:14:01 <ehird> 20:54:29 <Sgeo> I mean, it hardly makes sense to define it in this spec, since the way my product uses it will require more than just this spec
14:14:02 <ehird> Stop calling it a product, you're not a business.
14:14:18 <ehird> 21:16:38 <Sgeo> The product will allow people to set up positions where people who sit on furniture end up sitting
14:14:19 <ehird> Damn that's some useful product
14:31:07 <ehird> 02:09:00 <fizzie> Hrm, my embedded rfk86 font (after generating a SVG vector font out of the bitmaps, and converting that to truetype with fontforge) didn't turn up so great: http://zem.fi/~fis/ugly.png
14:38:34 <fizzie> It's not *that* horrible.
14:39:43 <Deewiant> Looks fine to me, apart from the obvious sidebar mess
14:41:13 <fizzie> The space glyph maybe got dropped, since it has no pixels of content.
14:41:56 <Deewiant> I doubt it got completely dropped, or those blank lines would have two ||s on the left with no space in between
14:42:53 <fizzie> I'm pretty happy that a simple "font-size: 12px;" got me a fixed *2 scaling.
14:43:05 <ehird> It looks nothing like the screenshots, at least.
14:43:34 <fizzie> Maybe it substitutes a space from somewhere, who knows.
14:43:47 <ehird> Well... do you have a space character?
14:44:10 <ehird> Is it distinct from the glyph at a codepoint you don't implement? (Maybe "no codepoint"=that)
14:44:34 <fizzie> Uh, it's the exact same font, with the same scaling than in the screenshots.
14:44:58 <ehird> fizzie: Well, okay then.
14:45:27 <ehird> Incidentally, who the hell uses Midori?
14:45:43 <fizzie> (using a phone, expect laggy replies)
14:46:58 <fizzie> It was the least-bytes webkit thing I saw in repo; epiphany wanted some 90M of Gnomish stuff.
14:47:16 <ehird> Arora is the most popular one, I gather.
14:47:24 <ehird> But hey, no knocking of Gnomish stuff allowed :P
14:48:09 <fizzie> I don't really use it. Ff3.5 should do embedded Truetype, but it didn't want to.
14:48:27 <ehird> It is a bad Firefoxy.
14:48:35 <ehird> You have to use a special font format, don't you?
14:48:44 <ehird> Maybe. I think that's the "widely"-supported thing.
14:49:06 -!- Pthing has quit (Remote closed the connection).
14:49:41 <fizzie> There were all kinds of gobject-introspection-whatevers I didn't have preinstalled.
14:50:44 <fizzie> The special format is for IE, or at least that's what one cross-browser embed page said.
14:51:17 <fizzie> Have to look more closely when I get home.
14:52:20 <fizzie> There was a small CSS trick to make it work simultaneously on IE and others.
14:58:34 -!- Pthing has joined.
15:01:07 * Ilari is fighting with pulseaudio.
15:04:05 <Ilari> Or actually, now with SDL...
15:21:36 -!- impomatic has joined.
15:21:41 <impomatic> Geocities sites will be removed after 26 October. :-(
15:22:00 <impomatic> Has the interesting Esoteric stuff been mirrored already?
15:23:19 <Ilari> Yay. Got sound to work. But mplayer no longer works (not related to pulseaudio)... :-/
15:24:20 <Slereah> Really, most don't exist anymore anyway
15:24:29 <Slereah> When's the last time you fund a functioning geocity site?
15:30:58 -!- fax has joined.
15:49:22 <Ilari> Oh yay, pulseaudio process has 230MiB vmsize... Apparently likes to mmap lots of stuff...
15:49:50 <Deewiant> Who cares about the virtual size?
15:50:23 <Deewiant> The wine processes are 2.5 gigs each
15:51:33 <Deewiant> (foobar2000.exe as well as the support processes, explorer.exe, winedevice.exe, services.exe)
15:58:26 <Ilari> Heh... Apparently pulseaudio has >200 megs of shared memory segments...
16:00:27 -!- kar8nga has joined.
16:06:48 <ehird> What Deewiant said, who cares about virtual memory?
16:06:59 <ehird> ((Unless you're using some UBER-LAME 32-BIT PROCESSOR or something))
16:14:05 <Ilari> Even with 32-bit, it is 3GiB of virtual memory per process.
16:45:13 -!- adam_d has joined.
16:58:44 -!- Asztal has joined.
17:25:24 -!- Sgeo has joined.
17:41:36 -!- oerjan has joined.
17:44:36 <oerjan> <ehird> Whoa, hi clog. <-- there is some serious time travel in the logs before that point...
17:47:08 <Sgeo> What does oerjan mean by time travel in the logs? The bot cutting out?
17:47:18 <oerjan> 04:17:39 --- names: list (clog BeholdMyGlory Pthing FireFly augur Gracenotes fungot coppro bsmntbombdood Slereah AnMaster lifthrasiir puzzlet Guest76637 Deewiant dbc MizardX ineiros sebbu2 fizzie SimonRC Gregor EgoBot rodgort Ilari pikhq Leonidas comex_ HackEgo mtve Warrigal olsner)
17:47:20 <fungot> oerjan: a wanderer i guess. especially since i took an english lit class
17:47:24 <oerjan> 03:54:27 --- log: started esoteric/09.10.17
17:47:32 <oerjan> those are two consecutive lines
17:47:42 -!- kar8nga has quit (Remote closed the connection).
17:49:38 <oerjan> although after all that, it is still at UTC -0700 afaict
17:50:53 <AnMaster> hm read that as "it is still an UTC -0700 addict"
17:53:34 <sebbu2> oerjan, you don't use ipot on a daily basis like everybody else ?
17:53:43 <oerjan> sebbu2: quoted line from the logs
17:54:14 <oerjan> i don't use apple products
17:54:40 <sebbu2> ( i have an old ipod, 40gb )
17:54:53 <sebbu2> ( the new ones have too little capacity storage )
17:54:57 * oerjan clearly assumed that was an internet-based drug from apple, based on the addict comment
17:55:31 <oerjan> i may be using ipot but i don't know what it is
17:57:35 <oerjan> sebbu2: the only relevant link seems to be in french
17:58:49 <sebbu2> oerjan, i'm french, and it was an april fool some year ( it was a protocol to transport ip over time, like accesssing a server in the past or in the future )
17:59:12 <sebbu2> like ipoac, except there is no real application
17:59:41 <sebbu2> ( ip over avian carriers )
18:00:12 <sebbu2> ( ipoac has already been used IRL )
18:01:02 * oerjan wonders if that recent south african experiment used ipoac but doubts it
18:04:51 -!- davidz has joined.
18:06:49 * Ilari chunks Pulseaudio into trashbin... The pulseaudio alsa driver is just plain too unstable.
18:07:11 <oerjan> it's just pulsating, silly
18:08:05 <Ilari> Deadlocks, wild CPU usage, assertion failures...
18:08:12 <davidz> hehe wrong channel. i need some moon freaks :D
18:09:34 <Ilari> And of course the pulseaudio binary is u+s (I did u-s to it) and is the classic for exploiting NULL pointer derefs.
18:10:05 <oerjan> davidz: one day we'll make the topic accurate again. at least for a few minutes.
18:10:43 <oerjan> but no, it's not about that kind of esotericism
18:11:11 <davidz> i cannot find those esoteric guys around here. it seems they don't like irc...
18:11:26 <oerjan> freenode may not be the best bet anyway
18:11:42 <davidz> does somebody in here want to beta test a moon app for hte iphone?
18:12:29 * oerjan already has a wall calendar with moon phases
18:13:11 <oerjan> and no iphone, for that matter
18:14:34 -!- davidz_ has joined.
18:15:58 -!- davidz_ has left (?).
18:19:46 -!- davidz has quit (Read error: 60 (Operation timed out)).
18:29:48 <ehird> wow, meaningless esoteric bullshit + trendy phone
18:29:51 <ehird> ironic and amusing
18:33:18 <Deewiant> "meaningless esoteric bullshit"? All he said was "moon freaks"
18:33:51 <AnMaster> well, I can't imagine why anyone would assume this was about moon stuff except that
18:34:00 <AnMaster> unless there is esoteric astronomy or something.
18:34:22 <oerjan> when the moon hits your eye like a big pizza pie *SPLAT*
18:34:33 <ehird> Deewiant: [18:11] davidz: i cannot find those esoteric guys around here. it seems they don't like irc...
18:34:36 <ehird> It's clearly about esooterica.
18:34:48 <ehird> Which, yes, is bullshit
18:35:47 -!- AnMaster has set topic: the world ends: ais523 has actually been thinking about Feather | http://tunes.org/~nef/logs/esoteric/?C=M;O=D | This is about esoteric programming languages..
18:36:09 <Deewiant> ehird: Being interested in that stuff doesn't mean you're full of bullshit.
18:36:19 <ehird> And you'll note that I never claimed that
18:36:33 <Deewiant> Your tone somewhat suggested it
18:36:33 <ehird> I wonder what kind of mouth e is.
18:36:49 <ehird> Deewiant: I was calling the thing stupid, not the person
18:37:16 <ehird> Although if you go to all the trouble of making an iPhone app for some esoterica there's a good chance that you do buy into the bullshit
18:37:24 <ehird> AnMaster: you are boring
18:37:54 <ehird> anyway, that sentence is totally false, for the majority of reasonable definitions of this
18:38:18 <AnMaster> I didn't say it was true right now
18:38:24 <ehird> I didn't mean that.
18:38:35 <AnMaster> though I have been coding on cfunge today
18:39:23 <ehird> I didn't mean that.
18:39:26 <AnMaster> code cleanup mostly. Dropping the boehm-gc support (it was slow, buggy and not well tested, plus caused quite a lot of messy code due to #ifdefs in various places)
18:39:35 <AnMaster> still only in a local branch though
18:39:44 <AnMaster> oh also working on integrating your cygwin fixes
18:40:17 <ehird> What was that fingerprint that didn't work in Cygwin?
18:40:40 <ehird> Also, why does boehm need ifdefs?
18:40:44 <AnMaster> ehird, as far as remember it was PERL wasn't it?
18:40:53 <ehird> No, it was TERM or something.
18:41:25 <AnMaster> anyway I have been working on integrating the other parts.
18:41:44 <AnMaster> but right now I'm trying to figure out how to manually compile bleeding edge gcc
18:41:55 <Sgeo> ehird, if I said that I'm writing specs for something, will you scream?
18:42:11 <ehird> No, because the chance that anyone else will read them is roughly nil.
18:42:42 <ehird> AnMaster: By the way, the getaddrinfo thing is only needed for IPv6, iirc.
18:42:47 <Sgeo> My competitors will read them if they want to make something compatible with my product
18:42:48 <ehird> Isn't there another function just for v4?
18:43:00 <ehird> Sgeo: Get over yourself, you're not a company and you don't have any competitors or products.
18:43:12 <AnMaster> ehird, hm iirc the alternative is deprecated or obsolete or something
18:43:23 <ehird> AnMaster: Yes, well, you could argue that Windows is too.
18:43:24 <Sgeo> of what you said is correct
18:44:21 <AnMaster> ehird, Deewiant, btw I hit an odd issue with mycology and/or ubuntu recently.
18:45:47 <AnMaster> I was not connected to any network (that is, out of wlan range, no ethernet), mycology tests in SCKE keept failing. Didn't have time to check more right then. But what exactly is it resolving?
18:46:10 <ehird> Is localhost in /etc/hosts?
18:46:20 <AnMaster> ehird, that was the *first* thing I checked
19:19:43 -!- Pthing has quit (Remote closed the connection).
20:12:19 -!- oklopol has joined.
20:12:48 <oklopol> helloes, my computer rebooted, and i was too lazy to reopen freenode
20:18:05 <oklopol> AnMaster: i've done some querying, and it seems limits aren't taught rigorously for all our computer science students either.
20:29:13 -!- Pthing has joined.
20:29:52 -!- oerjan has quit ("Good night").
20:37:27 -!- impomatic has quit ("mov.i #1,1").
20:46:49 <fizzie> Ha, well, that explains why the font embedding examples don't work in my Firefox: noscript.forbidFonts, default, boolean, true.
20:47:00 <fizzie> There's indeed a "forbid @font-face" checkbox set in the options.
20:47:59 <fizzie> Also explains why noscript had that "I've blocked some stuff" icon for the page even though I have no actual scripts there; I did wonder about that.
20:48:28 <ehird> I love how NoScript disables stuff that isn't a script in any way nor resembles one at all!
20:48:54 <fizzie> I'm sure there have been security problems with maliciously crafted truetype fonts.
20:49:14 <ehird> I very much doubt that. (Was that a joke?)
20:49:39 <fizzie> Here's the first google hit: http://xforce.iss.net/xforce/xfdb/47103
20:49:49 <fizzie> (Admittedly it's for the java runtime and does not really apply.)
20:50:28 <fizzie> Don't they have some sort of virtual-machine bytecode thing for hinting and all?
20:50:37 <fizzie> Everything like that has security bugs.
20:51:29 <fizzie> Here's another in FreeType 2: http://www.vuxml.org/freebsd/4fb43b2f-46a9-11dd-9d38-00163e000016.html
20:51:46 <fizzie> "one-byte heap-based buffer overflow via a specially crafted TTF file".
20:52:30 <fizzie> Besides, it resembles a script in the "2. handwriting, hand, script -- (something written by hand; "she recognized his handwriting"; "his hand was illegible")" sense. :p
20:56:18 <fizzie> Heh, in Ff it also converted "font-weight: bold;" sections into some really ugly blurry antialiased mess.
21:00:31 <ehird> Illegal object error.
21:03:26 <Asztal> the "forbid @font-face" must be pretty new, it only appeared when I updated just now
21:04:27 <fizzie> It also seems to be just the space glyph that has the wrong size; I have a <glyph /> tag in the SVG for the space with an empty d="" path argument, and loaded in fontforge that turns into a 'X' which I think means no-glyph; if I "modify" it by dragging the borders around, I get a blank square. I guess I need to put something in the glyph so that fontforge won't ignore it.
21:09:23 <fizzie> Yay, putting a "M0,0" dummy-path there "fixed" it; it's invisible, but it's still a glyph as far as fontforge is concerned. Maybe not the most elegant solution ever.
21:12:12 <ehird> Try deleting the path now >:)
21:12:46 <fizzie> There's still wrong line-spacing for the font; there are small gaps between the vertical line-drawing characters. (Even though the vertical advance should be 1em by default, and I have units-per-em="576" and the character polygons are 576 units high.
21:15:50 <ehird> You know, I'd just use one image per character and some JS to auto-do it rather than fight fontforge's insanity.
21:16:14 <ehird> Oh. I thought you'd stubbed a toe.
21:16:19 <AnMaster> full bootstrap, worked, except that it ignored my LDFLAGS in building stage3
21:16:30 <AnMaster> thus meaning it ignored -rpath
21:17:08 <AnMaster> which means I either have to use LD_LIBRARY_PATH every time I want to use it, or try to figure out how to make it not ignore that flag (pretty sure I managed it before even so...)
21:19:05 <AnMaster> coppro, what do you suggest instead?
21:19:20 <coppro> it's just generally not worth the effor
21:19:27 <fizzie> But I want to be a real font when I grow up. Strange, though; ascent+descent == em size, and the full-height characters are em-size units high. I guess something somewhere might be confused by my non-standard not-a-power-of-two em-size, but it's not very easy to find a integer N such that 2^N is evenly divisible by 6 (number of rows in the bitmaps).
21:19:31 <coppro> (although you could try clang if you're just using it for C)
21:19:37 <pikhq> Idea: murder their build system.
21:20:05 <pikhq> Why oh why do they not actually use autoconf/autotools (or a superior alternative)?
21:21:08 <fizzie> Heh, the easy fix: CSS line-height: 12px; -- it even works in this browser.
21:23:03 <AnMaster> <pikhq> Why oh why do they not actually use autoconf/autotools (or a superior alternative)? <-- what are they using then?
21:23:55 <pikhq> It's autoconf and custom Makefile.in, IIRC.
21:24:29 -!- MigoMipo has quit ("Page closed").
21:24:30 <AnMaster> really if you are going to use autoconf, use automake too.
21:24:45 <AnMaster> autoconf but no automake is way way way worse than autoconf+automake
21:25:11 <AnMaster> of course autoconf+automake is still a pain
21:25:29 <pikhq> Autoconf+automake is kinda bad, but at least it's consistent and understood...
21:25:52 <pikhq> Autoconf+custom Makefile.in makes baby Jesus cry.
21:26:18 <AnMaster> pikhq, consistent I agree about. But "understood"?
21:26:28 <AnMaster> how is custom one harder to understand as such
21:26:41 <AnMaster> except as a side effect of being inconsistent
21:26:52 <pikhq> All that's understood about autoconf+automake is how to deal with it.
21:26:53 <Sgeo> I once learned COBOL </completely-random>
21:27:06 <pikhq> The internal functioning is... Incomprehensible.
21:27:37 <AnMaster> pikhq, my point here was that the "less understood"-ing of autoconf+custom Makefile.in is just a side effect of the inconsistency.
21:28:02 <AnMaster> it is pretty easy to understand a Makefile.in you wrote yourself
21:28:42 <AnMaster> the thing is simply that configure substitutes @FOO@ with FOO defined using some autoconf macro
21:29:11 <AnMaster> Makefile.am differs a *lot* more from the final Makefile than Makefile.in does
21:29:39 <AnMaster> actually there might have been a conditional thing too in .in.. Analogous to #ifdef in C basically
21:30:02 <AnMaster> coppro, sure it isn't for automake only?
21:30:46 <pikhq> And then there's GNU Make's conditionals.
21:31:15 <pikhq> So many freaking layers to the GNU buildsystem...\
21:31:43 <coppro> I have been working with SCons recently and very much like it
21:32:10 <pikhq> SCons is kinda poor.
21:32:26 <AnMaster> cmake is quite nice apart from a slightly verbose syntax.
21:32:35 <pikhq> In that it's not a build system so much as it is a library with which you could reasonably *write* a build system.
21:32:37 <ehird> pikhq: why are you bothering to point that out to someone who doesn't dislike C++ :)
21:33:04 <ehird> coppro uses C++ and doesn't dislike it
21:33:21 <ehird> (he avoids the term "like" explicitly and I'm doing in kind, but it fools nobody :P)
21:33:28 <pikhq> coppro: The beatings shall continue until good taste evolves.
21:33:41 <coppro> pikhq: the thing is, though, that you can extend it however you want
21:33:53 <ehird> That could refer to both Scons and C++, and that scares me.
21:34:43 <ehird> Also: guys, the wired internet we can get here... will range from 3 to 4 megabits. The maximum speed of this 3G stick? 3.7 megabits. Yeaah... even the US is better than this in good places (erm, right?)
21:34:46 <coppro> SCons is entirely internally replaceable and extensible, other build systems make this difficult, in my experience
21:35:10 <ehird> in the old place i could get 800KiB/s out of an 8Mb (= 1024KiB/s) connection :(
21:35:25 <ehird> it doesn't help that we're on the exchange of the nearby town Prudhoe
21:35:28 <pikhq> coppro: The same is true of Autoconf, Cmake, custom shell scripts, custom Perl scripts, and even "To build, cc *.c -o foo".
21:35:30 <ehird> in hexham there was an exchange in town
21:35:58 <ehird> coppro: if you ever need or want to totally replace or extend it, it's broken
21:36:52 <coppro> pikhq: No. Anything based on Make directly or indirectly has limitations of Make's irritating dependency system
21:37:35 <ehird> I wish there was a dance with accompanying song going like: "working around the limitations of the language in tools, yeah yeah yeah" because I would dance and also sing it.
21:37:52 <pikhq> And anything based on SCons inherently has the limitations of IT BASICALLY BEING A CUSTOM PYTHON SCRIPT WHICH HAPPENS TO USE A LIBRARY FOR THE BUILDING.
21:38:15 <ehird> SCons: I'm certain we can make Python into a declarative language if we try hard enough.
21:38:15 <Sgeo> Should I learn Scala?
21:38:36 <ehird> I learnt it, liked it, realised it sucked, and yeah.
21:38:44 <pikhq> I will admit that Make is kinda poor. But SCons is a step backward from Autotools.
21:38:59 <ehird> (It sucks mainly because you still have 90% of the shit from Java pissing you off, but also because of some language flaws.)
21:39:03 <pikhq> (still better than Ant -- that's a step backward from Make.)
21:39:10 <ehird> pikhq: WHOA, I'm not sure I'd go that far (wrt scons/autotools)
21:39:31 <ehird> Let's just call them equals near the bottom of the good-build-system list.
21:40:06 <pikhq> ehird: But Scons isn't even a build system.
21:40:21 <ehird> Neither is autotools, it's actually an incarnation of Cthulhu.
21:41:20 <coppro> just out of curiosity, how do you define it as not being a build system?
21:41:44 <coppro> because you only have to run one program for it to work?
21:43:33 <pikhq> No, but because you are writing a somewhat involved Python script which happens to make a few calls to Scons at times.
21:44:21 <AnMaster> <ehird> in the old place i could get 800KiB/s out of an 8Mb (= 1024KiB/s) connection :( <-- that is quite good. Around what I get from mine
21:44:43 <ehird> It's not good compared to the 50Mb you can easily get elsewhere.
21:44:54 <AnMaster> ehird, elsewhere being big cities?
21:45:04 <ehird> No, e.g. Sweden/Finland/Norway.
21:45:17 <AnMaster> ehird, sorry, you can't get that in small towns
21:45:23 <AnMaster> and not over ADSL anywhere I bet
21:45:37 <AnMaster> isn't like 24 MBps the limit of ADSL?
21:45:53 <AnMaster> cable or fibre can give you more yes
21:45:57 <ehird> What you mean to say is "you can't get that in my small town".
21:46:23 <ehird> I know ADSL/ADSL2+, and you mean ADSL2+.
21:46:28 <ehird> No. There is fibre-optic.
21:46:35 <ehird> That is what the 50Mb connections are.
21:46:44 <AnMaster> ehird, well I was saying the limit of ADSL. With that I mean ADSL*
21:47:14 <ehird> ADSL is a worthless hack anyway. Fibre-optic is the only sane internet delivering system...
21:48:03 <ehird> If only companies would offer to charge you an insanely expensive set up fee, as opposed to just saying "nope, haven't laid pipes there yet".
21:48:27 <AnMaster> I think there are plans on some soft of fibre optic gird in this town
21:48:29 <pikhq> Also, "env = Environment(ENV = {'PATH' : os.environ['PATH']})" is criminal.
21:48:48 <ehird> We'll probably get the cheapest >4Mb connection we can find, just for the lower latency than 3G...
21:48:53 <ehird> ...but it's really, really annoying.
21:49:06 <AnMaster> pikhq, erhm... what is env after that
21:49:28 <AnMaster> because it seems we have set too many layers of PATH there
21:49:56 <ehird> You're making no sense.
21:50:17 <ehird> Ugh, why am I getting so tired these days?
21:50:32 <pikhq> AnMaster: That's creating an environment object which has PATH equal to the global environment's PATH.
21:50:34 <ehird> Exactly coinciding with moving, in fact.
21:51:07 <ehird> Maybe because this room has no curtain, and possibly I need a brighter light.
21:51:19 <AnMaster> env = Environment(ENV = {'PATH' : os.environ['PATH']})
21:51:29 <ehird> You fail at Python forever.
21:51:36 <AnMaster> ehird, oh yeah it is python. right
21:51:46 <ehird> "ENV =" is your slipup.
21:51:46 <AnMaster> then it makes a bit more sense
21:51:47 <pikhq> s/Set ENV to/the ENV argument is/
21:51:50 <ehird> btw, pikhq, it should be ENV=
21:52:10 <AnMaster> pikhq, but why. What other sort of argument could an environment have
21:52:27 <ehird> AnMaster: ENV = environment variables
21:52:30 <ehird> Other things = cflags, etc
21:52:35 <pikhq> AnMaster: CPPPATH,
21:52:41 <ehird> SCons::Environment != OS::Environment
21:52:47 <ehird> Scons::Environment contains OS::Environment
21:52:57 <AnMaster> cflags is also an environment variable to me, but whatever
21:53:12 <coppro> pikhq: how involved the python script is depends on what your building
21:53:22 <ehird> coppro: we've used SCons too, you know
21:53:26 <ehird> probably for longer than you
21:53:30 <ehird> we know how it works.
21:53:44 <coppro> nonetheless, I feel an urge to correct any generalizations
21:54:07 <ehird> I guess we're just misguided and don't have any valid complaints whatsoever.
21:54:31 <pikhq> coppro: env = Environment(ENV=os.environ) is already going beyond what should be needed...
21:54:53 <coppro> you don't have to do it, you know
21:56:02 <ehird> Technically true; also useless.
21:56:40 <pikhq> Do if you don't want to blithly ignore CC, CFLAGS, and friends...
21:57:32 <coppro> I'm not of the opinion such things should be environment vars... that said, I do wish it had make's command-line configuration ability
21:57:40 <coppro> (without you adding it, that is)
21:58:38 <pikhq> Explicit handling of CC and CFLAGS is a sign that you're doing it REALLY FUCKING WRONG.
21:58:46 <coppro> I don't think CFLAGS should be an env var, but I would like more scons CFLAGS=*bla*
21:59:53 <AnMaster> okay I just edited the generated Makefile of GCC in a hope that it would help...
22:00:12 <AnMaster> coppro, actually it seems to now. So far
22:00:45 * coppro kills the GCC build system with an electrified spork
22:00:57 <pikhq> coppro, you're not qualified to.
22:01:07 <AnMaster> scons here would be even worse
22:01:21 * coppro thinks for a second, then does the same thing to the entire source tree
22:01:27 <AnMaster> that would probably mean rewriting large parts of it just to get it to listen to LDFLAGS.
22:02:05 <AnMaster> ah wait no need to edit makefile
22:02:14 <AnMaster> I could override by setting those on command line of make
22:02:38 <AnMaster> make BOOT_LDFLAGS='...' LDFLAGS_FOR_TARGET='...'
22:03:57 <pikhq> http://www.scons.org/doc/production/HTML/scons-user/x4076.html ...
22:04:04 <AnMaster> also ... is replaced with -Wl,-rpath,/home/...
22:04:37 <AnMaster> pikhq, what do you think of cmake? Apart from the overly verbose syntax
22:05:01 <pikhq> AnMaster: Cmake is rather well-done.
22:05:14 <pikhq> And its syntax doesn't seem all that verbose...
22:05:14 <AnMaster> pikhq, not the syntax though. You can't claim that.
22:05:26 <pikhq> Though it apparently used to be.
22:05:40 <coppro> cmake is based on make :(
22:05:44 <pikhq> "if(foo) bar endif(foo)" is how they used to do it. :(
22:05:49 <AnMaster> coppro, what is wrong with that?
22:06:07 <coppro> but make sucks. It really does.
22:06:14 <pikhq> coppro: No, the make backend is merely one backend.
22:06:38 <AnMaster> <pikhq> "if(foo) bar endif(foo)" is how they used to do it. :( <-- used to? They dropped that?
22:06:39 <coppro> yeah, because I really want to make an IDE project :/
22:06:42 <pikhq> And you're a proponent of something that's less capable than Cmake. Shaddup.
22:06:46 <ehird> Can we stop talking about shitty compilers for shitty languages and their shitty tools?
22:06:56 <pikhq> AnMaster: "if(foo) bar endif()" now.
22:06:58 <ehird> Shut up, C++ user.
22:07:04 <pikhq> ehird: They're still relevant.
22:07:16 <ehird> Yes, but it's so upsetting.
22:07:20 <AnMaster> pikhq, hm. that () seems a bit out of place
22:07:22 <ehird> Or, you know, not.
22:07:25 <ehird> But it's mildly irritating!
22:07:44 <ehird> oklopol: your mom is irrelevant
22:08:03 <ehird> no, that was pretty terrible
22:08:21 <coppro> hmm... /me starts wondering whether cmake can be considered a build system
22:08:33 <ehird> ah, the classic zealot tactic
22:08:46 <pikhq> Okay, and this is the embodiment of evil: http://www.scons.org/doc/production/HTML/scons-user/x4105.html
22:08:51 <AnMaster> ehird, from the point of view of being a good insult yes. From the point of view of being funny, no
22:08:53 <ehird> "And now... I will search for tenuous ways to attack YOUR favoured system! Arguments? Fuck arguments, I'll attack YOUR preferences!"
22:09:06 <pikhq> "if not env.GetOption('clean')"? AAAAAAAAAGH!
22:09:13 <ehird> "Your build system was written by fags! QED. I win, you are incorrect."
22:09:33 <coppro> pikhq: no disagreement on that point
22:09:47 <ehird> Ooh, ways that is not pythonic: CamelCase. Saying "get" in a method name in that way, ever. Not using ['foo'] instead.
22:09:54 <AnMaster> is that supposed to be build rule?
22:09:55 <ehird> So, everything apart from the "env."!
22:10:04 <ehird> AnMaster: scons has no build rules
22:10:12 <ehird> in fact it's hard to say what it does have
22:10:14 <AnMaster> ehird, how does it do incremental rebuilds then
22:10:27 <ehird> Badly. But yes, it does them.
22:10:43 <coppro> AnMaster: you call functions of the environment to indicate you want to build a given target
22:10:45 <pikhq> The build commands for scons each only build if necessary.
22:10:49 <ehird> It basically shoehorns Python into a declarative language, and then adds a tiny imperative wrapper over your build declarations that are 99% imperative code.
22:10:58 <AnMaster> coppro, that doesn't even make sense in any way whatsoever
22:11:02 <ehird> (Except it's actually implemented even more shitty than that)
22:11:04 <AnMaster> ehird, what about out of source tree builds
22:11:11 <AnMaster> as in build tree != source tree
22:11:14 <ehird> AnMaster: What's that got to do with anything, yes it does that
22:11:17 <pikhq> But it is actually a freaking Python script and going all imperative, executing each one in order.
22:11:31 <AnMaster> ehird, ah. Was almost suspecting it would fail badly at that too
22:11:42 <coppro> it's actually very good at that one
22:12:08 <coppro> AnMaster: env.Program('program', ['source.c', 'more-source.c'])... it process all such calls, orders up the dependencies, and then executes them
22:12:12 <ehird> The "out of tree part", perhaps. The "build" part, no.
22:12:18 <ehird> Both are required to be good at out of tree builds.
22:12:34 <pikhq> In a way, I wish that someone would replace make with something that does exactly what make does, but better. Nuking such stupid "features" as "not supporting spaces in filenames" and "not knowing WTF quoting is"...
22:12:47 <ehird> Seriously, it's probably best to stop defending SCons, coppro; our collective knowledge of it almost certainly exceeds yours.
22:13:06 <coppro> pikhq: I do not disagree with that
22:13:32 <AnMaster> pikhq, um anyone using a space in a filename of a source file should be shot anyway
22:13:35 <ehird> Kind of like how you don't dislike C++ eh
22:13:48 <ehird> Joke, see, it was a joke.
22:13:56 <ehird> AnMaster: Why? Is this the 80s? Also, since when is make only for code?
22:14:18 <pikhq> I tried using make to do some massive format conversion on some media files.
22:14:22 <ehird> Personally I'd love to have an ecosystem where we can name something what it is without concern for stupid backwards-compatibility that's actually just with tools, not the underlying system.
22:14:25 <AnMaster> ehird, because typing $EDITOR "foo bar.c" is more work than $EDITOR foo_bar.c
22:14:30 <coppro> another "feature" to nuke is the total lack of any ability to generate dependencies dynamically
22:14:40 <oklopol> AnMaster: anyone who thinks spaces shouldn't be used in filenames should get a better os.
22:14:42 <AnMaster> <pikhq> I tried using make to do some massive format conversion on some media files. <-- Um... You are doing it wrong
22:14:46 <ehird> You can have a target that generates a Makefile
22:14:50 <AnMaster> that just isn't what make is meant for
22:14:56 <ehird> It was not meant as a build system
22:14:59 <pikhq> AnMaster: Yes it is.
22:15:16 <coppro> ehird: was just about to say that doesn't count
22:15:18 <pikhq> "%.ogg: %.flac" -- This should just plain work.
22:15:19 <AnMaster> ehird, oh really? Interesting. Care to link me to it?
22:15:21 <coppro> since it requires two passes
22:15:29 <ehird> AnMaster: Nope, you can dig yourself out of the hole
22:15:36 <ehird> coppro: Oh come on; tenuous. It's behind-the-scenes, it automatically reloads.
22:15:51 <ehird> AnMaster: But FYI, it was intended to transform any sort of file into another
22:16:02 <ehird> (And didn't give any thought to "projects", thus why using it as a build system is such a bitch)
22:16:05 <pikhq> http://en.wikipedia.org/wiki/Mk_(software)
22:16:46 <coppro> ehird: but not at all useful... for instance, you have to have those files exist (as blank files) before you generate them.
22:16:49 <ehird> I just use objects that use other objects and never need to build, what about you guys? Don't you use ehirdOS? :-)
22:17:29 <ehird> It's... simply false.
22:17:34 <ehird> Opposite of truth.
22:17:51 <oklopol> coppro: ehird does not explain, he mocks
22:18:05 <ehird> oklopol: shut up idiot
22:18:08 <pikhq> coppro: You don't have to have those files exist before you generate them.
22:18:11 <ehird> coppro: It just isn't true.
22:18:12 <pikhq> That's all there is to it.
22:18:15 <ehird> There's no element of truth to it.
22:18:25 <coppro> pikhq: in my experience, make errors out if you try to use another makefile that doesn't exist
22:18:26 <ehird> How can we explain anything that's simply plainly, boringly false?
22:18:40 <ehird> coppro: ...that's totally unrelated to anything we've said
22:18:44 <ehird> I meant a target with the filename "Makefile"
22:18:45 <pikhq> No, it warns. And attempts to make the file you included.
22:18:49 <ehird> Read the GNU Make manual to find out what it doeses
22:18:52 <ehird> But what pikhq said too
22:19:06 <coppro> was thinking -MD style generated includes
22:19:08 <pikhq> It errors out if there's no rule to build that file.
22:19:22 <coppro> that happen as a side effect of other compilations
22:19:37 <pikhq> "include foo.d bar.d baz.d" in a Makefile implies the rule "Makefile: foo.d bar.d baz.d".
22:20:07 <ehird> coppro: HOW DO YOU EXPLAIN THE FACT THAT I JUST EXPLAINED THINGS :P
22:20:39 <pikhq> What you do is create a rule for "%.d: %.c", and you've got automagic dependencies.
22:20:42 <oklopol> ehird: i bet you just did it to mock me
22:20:46 <pikhq> (shame that's not an implicit rule)
22:20:53 <ehird> oklopol: i did it totally accidentally in fact
22:21:02 <ehird> pikhq: huh, I never thought of that
22:21:11 <coppro> pikhq: ah, I suppose that would work
22:21:14 <pikhq> ehird: In the GNU Make info page.
22:21:18 <ehird> I always find myself overriding the default rules because there's spaces for variables I didn't use
22:21:22 <ehird> and they annoy me :)
22:21:38 <ehird> You can get it as HTLM.
22:21:43 <ehird> Just google "gnu make manual"
22:21:57 <oklopol> ehird: too tired to think of an answer
22:22:30 <AnMaster> info isn't that bad. Compared to, for example, .chm
22:22:50 <AnMaster> virtualbox comes with documentation in .chm format
22:22:51 <ehird> Hey, don't knock .chm too much.
22:22:59 * coppro fiddles around some more trying to get make to handle precompiled headers in a sane manner
22:23:01 <AnMaster> so you need to have kchmviewer installed
22:23:06 <ehird> You're probably mocking it because it's a Windows thing, but it's a fairly well-designed hypertext system.
22:23:16 <ehird> info is *at least* 100 times worse.
22:23:20 <AnMaster> ehird, oh what about old .hlp then?
22:23:33 <ehird> Pretty similar, really.
22:23:34 <pikhq> Info's a bit of a hack.
22:23:48 <pikhq> It at least works decently, but... Yeah, it's a hack.
22:23:53 <pikhq> GNU seems to do a lot of hacks.
22:24:01 <coppro> you know what the best (not) thing about make and precompiled headers is? I can't get it to work because GCC is broken.
22:24:13 <ehird> info does not work decently
22:24:19 <ehird> every time I run it I quit without getting a solutioino
22:24:20 <AnMaster> pikhq, well info syntax isn't quite as bad as man page syntax IMO
22:24:23 <ehird> it's painful to navigate
22:24:24 <coppro> info is the emacs version of man
22:24:36 <coppro> except without the usefulness
22:24:36 <ehird> that analogy doesn't even make sense
22:24:44 <pikhq> No, info is a hypertext system.
22:24:45 <AnMaster> sure the default info application is horrible
22:24:45 <ehird> emacs simple editing = vim simple editing in usage
22:24:54 <ehird> info(1) is unusable, unlike man(1)
22:24:56 <coppro> *ehird: lots of needless complexity
22:24:58 <ehird> in even its most basic usage
22:25:03 <pikhq> Man is not hypertext, it's just text.
22:25:06 <AnMaster> ehird, you know that emacs has a built info reader right?
22:25:07 <ehird> AnMaster: info is the GNU software
22:25:13 <ehird> we are talking about it, as well as the underlying format
22:25:15 <ehird> Yes, I do, it's bad
22:25:16 <coppro> ehird: was comparing emacs to a plain text editor, not to vim, of course
22:25:22 <ehird> Worse than man(1), like every info reader
22:25:46 <coppro> man isn't itself a reader
22:25:52 <AnMaster> ehird, yes that I agree about, but man doesn't work well for large stuff, like emacs docs. Imagine all the documentation of emacs in a single man page
22:26:07 <pikhq> Man is a shell script that does nroff with appropriate options and pipes it into PAGER.
22:26:20 * ehird views ls(1) in Xcode, which converts it to HTML first, just to annoy a large section of the participants in the current discussion
22:26:33 <ehird> mac haters, gui haters, ide haters, man haters, html ubiquity haters
22:26:46 <AnMaster> /usr/bin/man: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.15, stripped
22:26:49 <coppro> discussion? WTF! I'm leaving
22:26:55 <pikhq> Really, it's not a shell script?
22:27:12 <ehird> I have a BSD handy!
22:27:15 <pikhq> ... So it's a *program* that does system(nroff) and system(PAGER)...
22:27:27 <ehird> [~]$ file $(which man)
22:27:28 <AnMaster> * ehird views ls(1) in Xcode, which converts it to HTML first, just to annoy a large section of the participants in the current discussion <-- konqueror man:/
22:27:31 <coppro> ehird: I'm getting annoyed at the wrong part of your sentence. It's supposed to be slightly funny
22:27:41 <ehird> /usr/bin/man: Mach-O universal binary with 2 architectures
22:27:41 <ehird> /usr/bin/man (for architecture i386):Mach-O executable i386
22:27:41 <ehird> /usr/bin/man (for architecture ppc7400):Mach-O executable ppc
22:27:50 <AnMaster> ehird, of course firefox or nautilus can't do it
22:27:51 <Deewiant> Even on Solaris: /usr/bin/man: ELF 32-bit MSB executable SPARC Version 1, dynamically linked, stripped
22:28:04 <ehird> AnMaster: Firefox can open apt:pkg, I'm sure it could handle man:
22:28:20 <ehird> AnMaster: Anyway, that fails to annoy mac haters and ide haters
22:28:37 <ehird> Yes, there will be KDE haters and Konqueror haters, but mine annoys a probably-larger subset
22:28:51 <AnMaster> ehird, about apt, is just a protocol handler calling an external program
22:29:00 <AnMaster> ehird, man:/ in konq renders it nicely inside konq
22:29:04 <ehird> Deewiant: Several questions are raised, such as "why god why do you have a Solaris system to hand"
22:29:07 <ehird> AnMaster: Eh, whatever
22:29:10 <AnMaster> and it works fairly well for them too
22:29:21 <Deewiant> ehird: Because my school has them.
22:29:23 <ehird> Xcode doesn't add any styling to the man page so it's pretty ugly
22:29:43 <ehird> Although you can navigate to a section in the manpage in the little toolbar above it, which amuses me
22:29:46 <ehird> Someone coded that!
22:29:59 <ehird> (Although it probably just does it for every HTML page, since all the OS X docs are HTML)
22:30:12 <pikhq> Konqueror uses the same stylesheet as it uses for KDE's HTML documentation, I think.
22:31:13 <AnMaster> yay managed to build gcc correctly
22:31:57 <coppro> no such thing as building gcc correctly :P
22:32:01 <AnMaster> and it compiles cfunge correctly
22:32:09 <Deewiant> coppro: I wonder how Gentoo gets by?
22:32:28 <AnMaster> Deewiant, gentoo even offers installing multiple versions side by side easily
22:32:42 <AnMaster> to make sure I still stay compatible with older compilers
22:32:52 <ehird> you know, I'm fairly sure that the correct response to bad jokes isn't to take them seriously
22:32:56 <AnMaster> coppro, do they offer switching default system version too?
22:33:03 <ehird> because they'll just think you didn't get it
22:33:21 <AnMaster> ehird, actually I thought it was a bad insult
22:33:34 <ehird> Same thing; obviously there is a proper way to build gcc.
22:33:40 <ehird> The joke and insult is in suggesting that you can't.
22:34:08 <coppro> /usr/bin/cc is a symlink to /etc/alternatives/cc
22:34:51 <coppro> though it does not have alternatives for gcc
22:34:54 <ehird> [~]$ ls -l $(which cc)
22:34:55 <ehird> lrwxr-xr-x 1 root wheel 7 12 Feb 2009 /usr/bin/cc -> gcc-4.0
22:35:08 <ehird> [~]$ ls -l $(which c99)
22:35:09 <ehird> -r-xr-xr-x 1 root wheel 42816 27 Jun 2008 /usr/bin/c99
22:35:16 <ehird> AnMaster: Yes, I'm using Leopard
22:35:21 <ehird> In Snow Leopard, most shit uses LLVM/Clang
22:35:26 <AnMaster> and 4.0 is so buggy that I don't even bother testing with it
22:35:26 <Deewiant> /bin/cc: a /usr/bin/perl -w script text executable
22:35:33 <ehird> [~]$ ls -l $(which gcc-4.2)
22:35:33 <ehird> -rwxr-xr-x 1 root wheel 105680 1 Oct 2008 /usr/bin/gcc-4.2
22:35:41 <ehird> I bet Snow Leopard has a newer version, too
22:35:43 <AnMaster> ehird, what exactly is your /usr/bin/c99
22:35:57 <ehird> [~]$ c99 --version
22:35:57 <ehird> c99: illegal option -- -
22:35:58 <ehird> usage: c99 [-cEgs] [-D name[=value]] [-I directory] ... [-L directory] ...
22:35:58 <ehird> [-o outfile] [-O optlevel] [-U name]... [-W 64] operand ...
22:36:02 <ehird> AnMaster: POSIX-compliant thing, I guess.
22:36:04 <ehird> It'll just call gcc.
22:36:22 <ehird> C99(1) BSD General Commands Manual C99(1)
22:36:23 <ehird> c99 -- standard C language compiler
22:36:31 <ehird> The c99 utility interface conforms to IEEE Std 1003.1-2001 (``POSIX.1'').
22:36:53 <AnMaster> exec gcc -std=c99 -pedantic -U_FORTIFY_SOURCE ${1+"$@"}
22:37:00 <AnMaster> why on earth -U_FORTIFY_SOURCE
22:37:07 <ehird> For additional fortification!
22:37:34 <ehird> Well, the extra fortification must kill it, I guess.
22:37:36 <AnMaster> what reason is there to undefine that one.
22:37:51 <ehird> Maybe forts are bad.
22:38:57 <AnMaster> ehird, while I'm aware of that you are just being silly (and not funny) atm, do you actually know what _FORTIFY_SOURCE does?
22:39:57 <AnMaster> ehird, add overflow checks, so that stuff like sprintf with a too small buffer (if the size of the buffer is known statically) fails with an abort rather than results in possibly bad stuff happening
22:40:07 <AnMaster> for a number of standard functions
22:40:36 <ehird> Well, that's not C99-compliant, then.
22:40:54 <AnMaster> ehird, depends on if it is =1 or =2 iirc
22:41:05 <AnMaster> it only does build time checks at =1 level iirc
22:41:07 <ehird> Well, then it disables it just to be safe.
22:41:48 * AnMaster waits for ehird's reaction, guessing he never tried it before
22:42:06 <ehird> Ow ow ow MY SPLEEN
22:42:14 <ehird> It's so ugly it hurts my spleen
22:42:23 <coppro> here, have a mojo filter
22:42:30 <ehird> My eyes just went into override mode and ignored it entirely
22:42:32 <ehird> But my spleen, oh god
22:42:36 <AnMaster> it defines defaults for flags and parameters iirc
22:42:37 <ehird> My spleen cannot help but see
22:42:39 <AnMaster> and how some of them are handled
22:42:55 <AnMaster> though I'm not actually 100% sure if it does something else too
22:43:06 <AnMaster> %{mno-intel-syntax:-masm=att %n`-mno-intel-syntax' is deprecated. Use `-masm=att' instead.
22:43:16 <AnMaster> seems to indicate it can also warn about deprecated flags
22:43:24 <AnMaster> %{!mtune=*:%<mtune=native %:local_cpu_detect(tune)}} %{mtune=native:%<mtune=native %:local_cpu_detect(tune)}
22:43:39 <AnMaster> seems to implement the -march=native and -mtune=native stuff
22:43:46 <AnMaster> which is quite a weird place to put that
22:44:21 <AnMaster> %{march=native:%<march=native %:local_cpu_detect(arch) %{!mtune=*:%<mtune=native %:local_cpu_detect(tune)}} %{mtune=native:%<mtune=native %:local_cpu_detect(tune)}
22:45:07 <AnMaster> ehird, actually it reminds me of oil syntax
22:46:08 <AnMaster> -print-multi-lib Display the mapping between command line options and
22:46:08 <AnMaster> multiple library search directories
22:46:52 <ehird> I'm currently missing four keys and one doesn't press very well...
22:47:23 <ehird> I'm not sure you can return a keyboard for "I picked off the keys and can't get them back on again".
22:47:31 <ehird> [~]$ gcc -print-multi-lib
22:48:16 <Ilari> IIRC, one key once did get picked from this keyboard (then I figured how to push it back to its place).
22:50:04 <AnMaster> ehird, hm is library directory suffix for 64 bit libraries x86_64?
22:50:21 <ehird> *ahem* Time to recite my usual spiel. When I Get A Real Fucking Keyboard, Keycaps Will Indeed Slide Back On Verily So Easily.
22:50:25 <AnMaster> if not, my theory of what it means was wrong
22:50:32 <ehird> Also, The Sheer Immense Cost Will Probably Instill The Fear Of God Into Me
22:50:35 <AnMaster> ehird, why remove them all the time?
22:50:35 <ehird> AnMaster: Dunno. Maybe.
22:50:50 <ehird> I don't even notice myself doing it.
22:51:22 <AnMaster> ehird, bad habit. Try to break it (instead of the keyboard)
22:51:33 <ehird> Wow that's some helpful advice.
22:51:41 <ehird> "Why are you doing that?" "It's a habit." "Stop doing that."
22:52:34 <ehird> Therapy? For pulling keycaps off?
22:52:54 <ehird> "What are your hidden desires in pulling these keycaps off?"
22:53:01 <AnMaster> ehird, a) yes, b) yes and c) no
22:53:10 <ehird> "Are you unsatisfied with your life, perhaps, so that it gives you an... outlet into your desires, perhaps?"
22:53:13 <fizzie> Your search - "pulling keycaps off" therapy - did not match any documents.
22:53:19 <ehird> "A way to feel focused... as if you're moving on... perhaps?"
22:53:31 <AnMaster> fizzie, it will soon if google gets around to indexing this
22:53:41 <ehird> Google indexes us poorly, unfortuantely.
22:54:01 <ehird> Someday I'll write that awesome log-syncer script...
22:54:19 <ehird> Such goldmineosity contained twixt the logs of oor twain!
22:54:42 <fizzie> I got a perfect (well, in the "works as supposed" sense; it's still ugly and unreadable) font-embedding thing when seen through this Firefox now; but OS X's TrueType renderizer breaks the upper parts of the glyphs, for some curious reason.
22:54:47 <AnMaster> <ehird> Someday I'll write that awesome log-syncer script... <-- oh? what would it do?
22:55:29 <ehird> Disk. Download logs for the days you don't have (so you can keep up-to-date), don't download today's log (it's not complete yet), and convert all the times to UTC.
22:55:33 <ehird> Or local time, instead of UTC.
22:55:45 <ehird> That last one's a bitch, because you have to move messages between files and keep the start/end logging messages it puts.
22:56:05 <ehird> Oh, and maybe convert it to a slightly more readable format.
22:56:08 <ehird> Word-wrap it, for instance.
22:56:33 <AnMaster> ehird, just resize your editor and have it do it?
22:56:38 <ehird> (80 is a bit too wide for reading; 72 is what's mostly recommended for comments in code.)
22:56:41 <ehird> AnMaster: Every time?
22:56:46 <fizzie> AnMaster: Want to act as a lab test animal? Does the http://zem.fi/rfk86/screens.html page-text render properly and in the same font as in the screenshots?
22:56:46 <ehird> I edit non-log files too, you know.
22:56:49 <ehird> Anyway, it'd indent the stuff.
22:57:00 <ehird> Which most editors don't do when wrapping.
22:57:01 <Ilari> To get reliable timestamps, one should use UTC, TAI or POSIX time (with leap second markers) anyway...
22:57:06 <ehird> fizzie: Testing in Safari.
22:57:13 <ehird> fizzie: Looks fine, apart from the antialiasing.
22:57:17 <ehird> Which just makes it look smudgy.
22:57:27 <AnMaster> fizzie, what should render the same you said?
22:57:32 <ehird> fizzie: If you name your font, uh, what was it again? "Aargh" or something, Safari will disable antialiasing.
22:57:38 <ehird> (It's a hack because Acid 3 depends on the antialiasing behaviour)
22:58:12 <ehird> @font-face { font-family: "AcidAhemTest"; src: url(font.ttf); }
22:58:12 <ehird> fizzie: Call the font "Ahem".
22:58:17 <ehird> And maybe give it font-family AcidAhemTest, just to be sure.
22:58:26 <ehird> (Yes, this is an awful hack; isn't it wonderful?)
22:58:29 <AnMaster> fizzie, not on either of my computers. Tried the browsers I usually use on them (konq 3.x on desktop, firefox <whatever is default in jaunty> on the laptop)
22:58:53 <ehird> AnMaster: Firefox 3.0 doesn't support it.
22:59:01 <ehird> Konqueror doesn't support anything.
22:59:13 <AnMaster> ehird, konq uses much less ram
22:59:13 <ehird> fizzie: Are those screenshots meant to descend slightly, by the way?
22:59:27 <ehird> AnMaster: Use Epiphany or Arora or some other WebKit browser.
22:59:32 <fizzie> ehird: On this Ff the font-size: 12px; makes the font "pixels" correspond to 2x2 screen-pixels so that antialiasing doesn't cause it to go all blurry; but on my OS X (both Safari and Ff) the upper parts of the glyphs get curious |\/\/| style messups. Maybe some 10.4 bug/"feature".
22:59:35 <AnMaster> <ehird> fizzie: Are those screenshots meant to descend slightly, by the way? <-- they do in my firefox, but not in konq
22:59:54 <ehird> fizzie: It's regular antialiasing here, just makes it blur a bit.
23:00:00 <ehird> fizzie: You should just do the hack I said; it's quite trivial.
23:00:00 <AnMaster> ehird, you miss out on the nice kparts stuff
23:00:04 <ehird> Also, it's evil fun, which is nice.
23:00:18 <ehird> AnMaster: Mozplugger? Admittedly it's argh, but.
23:00:34 <ehird> AnMaster: Or, switch to OS X and use Safari. Best PDF-embedding plugin ever. Admittedly it uses RAM like an unholy beast...
23:00:38 <AnMaster> ehird, yeah I tried it. argh indeed. And that is firefox, not one of those webkit ones
23:00:58 <fizzie> ehird: I'll consider it. Anyway, it does the same blurry antialiasing here whenever the sizes don't match exactly, but when they do there's of course nothing to antialias.
23:00:58 <ehird> fizzie: I'm too lazy to do it myself but I must see! Dilemma.
23:00:58 <AnMaster> and OS X on a non-mac? hah hah. From what I heard that is a pain (though possible=
23:01:07 <ehird> I didn't say on a non-Mac.
23:01:13 <ehird> (Also, yes, it's possible; it's probably illegal.)
23:01:20 <ehird> You pay to buy OS X, too.
23:01:36 <ehird> Unless you meant pirate it and run it on a PC in which case, well, Steve Jobs will probably shit on your grave.
23:02:04 <ehird> Anyway, Safari isn't so bad... 770MiB of RAM used, yes, but I have something like 7 windows open with multiple tabs in each, mostly heavy pages.
23:02:13 <ehird> I open an awful lot of pages.
23:02:23 <fizzie> AnMaster: I just stuck them as float: left;s, saw the descending thing, decided it looks good enough. It wasn't an intended effect, so I guess not-descending is all right too.
23:02:26 <ehird> 11 windows actually./
23:02:32 <fizzie> http://zem.fi/~fis/foo.png is what happens on my Ff3.5.
23:02:49 <AnMaster> hm can't you disable the antialias with CSS?
23:02:51 <fax> looks nice
23:03:01 <fax> is this an emulator you wrote?
23:03:04 <AnMaster> and why on earth would acid3 depend on that
23:03:10 <ehird> 42 tabs in total, I counted.
23:03:15 <fizzie> fax: No, it's just a port of that game to the calculator.
23:03:16 <AnMaster> the workaround is a bad idea simply
23:03:21 <ehird> AnMaster: You can't disable it because it's a text thing. And to test pixel alighnment.
23:03:37 <ehird> People make a big deal of Acid 3, so. And it's an unintentional side-effect; they shouldn't really use the test.
23:03:47 <ehird> But Ian Hixie gave the go-ahead on it because it isn't really Safari's fault at all.
23:03:53 <ehird> (the author of acid 3)
23:04:10 <fax> Conclusions
23:04:10 <fax> OpenTTD is not really a very good platform to simulate digital logic circuits on
23:04:32 <fizzie> That conclusion has stood the test of time: I still think it's true.
23:04:47 <ehird> I wonder if FontForge has a binary.
23:04:55 <ehird> I wanna know if that hack still actually works...
23:05:05 <AnMaster> /usr/bin/fontforge: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.9, stripped
23:05:15 <ehird> *an OS X binary for download from the site
23:05:20 <ehird> AnMaster: Because, you know, it's quite obvious. :P
23:05:55 <AnMaster> ehird, anyway compiling it isn't hard (on gentoo at least)
23:06:14 <ehird> Yes, but that's too much effort just to see if fizzie could, in fact, find glory and wealth by using that hack.
23:06:20 <ehird> It is a fun hack though!
23:07:03 <AnMaster> glory and wealth by compiling fontforge for OS X?
23:07:12 <ehird> No, by naming the font Ahem
23:07:37 <fizzie> As far as I understand, for TTF-embedding whatever you write as the font-family: in the @font-face tag is what it uses, and it doesn't really care about the TTF contents at all; maybe it'd work simply by changing that.
23:07:38 <ehird> fizzie: Should you really have that exit screenshot there? It's such a spoiler!
23:07:41 <ehird> AnMaster: Then I checked, and corrected.
23:08:00 <ehird> fizzie: Then call it
23:08:22 <ehird> I could try it here if it's so simple, I guess.
23:08:33 <fizzie> But that breaks if the user has AcidAhemTest installed locally. :p
23:08:48 <ehird> AcidAhemTest, anotherfontname
23:08:53 <ehird> And have two @font-faces, one for each
23:08:58 <ehird> fizzie: But the font itself is called Ahem
23:09:00 <ehird> so there's no worry there
23:09:11 <ehird> It's just called AcidAhemTest in the @font-face
23:09:13 <fizzie> Actually I guess it doesn't unless you put AcidAhemTest also in the "local()" statement there, right.
23:09:15 <ehird> At least the last time I checked
23:09:56 <AnMaster> fizzie, what does the local() do?
23:10:14 <ehird> Make it check for a local font before downloading it?
23:10:15 <fizzie> AnMaster: Makes it use a locally installed font of that name if one exists.
23:10:33 <fizzie> ehird: As a side effect, it confuses IE, which would otherwise try to load the .ttf file too and fail.
23:10:35 <AnMaster> fizzie, why though? It isn't as if it is likely to exist elsewhere
23:10:44 <fizzie> AnMaster: For that side effect I mentioned.
23:10:46 <ehird> AnMaster: font-family: Zapfino
23:10:51 <AnMaster> fizzie, this feels like being back during IE 6 time
23:10:53 <ehird> Windows doesn't have Zapfino, for instance
23:11:02 <ehird> OS X would load it instantly
23:11:03 <AnMaster> <ehird> AnMaster: font-family: Zapfino, Fallback, Fallback
23:11:06 <ehird> The others would download it
23:11:16 <ehird> AnMaster: The whole point of @font-face is that you can use any font...
23:11:26 <AnMaster> fizzie, how do you generate the eot file?
23:11:43 <fizzie> AnMaster: There's that ttf2eot tool, hosted in code.google.com.
23:12:06 <ehird> eot is just a ttf iirc.
23:12:09 <ehird> Maybe in some wrapper.
23:12:27 <fizzie> There's something special, but I'm not sure what the special is.
23:12:46 <ehird> Anyway, change that one line in the CSS, you bum!
23:12:52 <fizzie> There's the binary-only Microsoft too (WEFT, Web Embedding Fonts Tool) that is the official way to generate EOT files.
23:12:56 <ehird> Technically it's "AcidAhemTest" with the quotes, but I doubt that's checked for
23:13:25 <fizzie> ehird: Okay, if it means so much to you, but you must then also test it with your OS X death-station.
23:13:48 <ehird> Also: As far as I know I have never stationed any deaths with this OS
23:14:33 <fizzie> Okay, now it should be "AcidAhemTest" in the CSS rules. Of course if it actually tests for the font name, that won't help.
23:14:38 <Sgeo> ehird, um, what?
23:14:57 <ehird> Is it just me, or does Sgeo never say what to anything that's actually not understandable in any way?
23:15:01 <ehird> It's always the plain, simple statements.
23:15:15 <Sgeo> "Also: As far as I know I have never stationed any deaths with this OS" What, exactly, does that mean?
23:15:36 <ehird> Are you ignoring fizzie?
23:16:18 <ehird> fizzie: Alas, the same. I don't suppose you could rename the font Ahem?
23:16:54 <ehird> "So Hyatt and I came to a deal. I would move the test down and to the left one pixel, so it doesn't affect the border anymore, he would accept to remove the hack, and would fix one additional bug (a background-position rounding bug)."
23:16:57 <ehird> — http://ln.hixie.ch/?start=1206756775&count=1
23:17:06 <ehird> It's worth a try, it may still be in there
23:17:06 <AnMaster> btw gcc 4.5 finally has true LTO
23:17:08 <ehird> But what sad news!
23:17:21 <AnMaster> or will have. was snapshot I built before here
23:17:24 <Sgeo> Wait, someone changed a test so that a browser would work with it?
23:17:35 <ehird> The test was wrong, basically.
23:17:38 <ehird> Or at least tenuous.
23:17:41 <ehird> Safari hacked around it.
23:17:48 <ehird> Then the Acid 3 author made that deal.
23:17:56 <ehird> It's not as simple an issue as you might think.
23:19:23 <Sgeo> What does it mean that the blue doesn't become blue on Acid3
23:20:18 <fizzie> ehird: Well, I changed the three names in fontforge's "names" tab, and uploaded as rfk86.ttf back. You might need to do some trickery to invalidate caches or whatnot, and it might still not be enough.
23:22:04 <ehird> fizzie: Seems like that code path isn't entered any more. In the mood for an awful hack? All OS X systems (that haven't been tinkered with) disable antialiasing for 4pt and below.
23:22:17 <ehird> So if you made the glyphs really ridiculously huge and set it to 3pt in an OS X-only conditional block (say with JS)...
23:22:26 <ehird> ...you'd go insane from writing a hack for such atrivial thing.
23:23:53 <fizzie> Right; I'll consider OS X -only fixups at some other point. Maybe I could get the same effect with enough twiddling of sizes and offsets somewhere; after all, the data is just rectangularish polygons, I'm sure OS X's anti-alias algorithm would also give sharp edges if I just got the polygons to match screen pixel borders.
23:24:36 <AnMaster> ehird, "All OS X systems (that haven't been tinkered with) disable antialiasing for 4pt and below." <-- why?
23:24:52 <ehird> "Turn off text smoothing for sizes [ 4 ↓ ] and below."
23:25:05 <AnMaster> ehird, err yeah but why that limit
23:25:23 <ehird> And because antialiasing text *that* small would just be a smudge; at least there's a semi-decent chance that any font that would be set that small is pixelly enough to be readable with a magnifying glass.
23:33:32 <ehird> Ehh, I really need to figure out how to make a serifed, italic pixel font.
23:42:05 <AnMaster> why do fonts on windows has such strange filenames
23:42:26 <AnMaster> it is quite common looking there
23:44:12 <Sgeo> So Ahem is unusable for normal text?
23:45:25 <ehird> I have no idea how to respond to that question... I'm serious.
23:45:34 <ehird> I have no idea how I could articulate the idea of why that's a ridiculous question to you.