00:09:32 -!- Asztal^_^ has quit (".").
00:09:34 <ais523> sometimes I think about Underload instruction minimisation
00:09:39 <ais523> in much the same way as BF instruction minimisation
00:10:57 -!- BeholdMyGlory has quit (Remote closed the connection).
00:11:23 <oerjan> : cannot be dropped, nor can * or ^
00:12:15 <oerjan> well, as far as being expressed by the others is concerned, at least
00:13:01 <oerjan> ! seems unlikely. maybe you could manage to sweep things down instead...
00:13:27 <oerjan> i don't think any of them are redundant
00:15:53 -!- ais523 has quit (Remote closed the connection).
00:18:19 -!- ehird has joined.
00:18:25 <ehird> oerjan: () is redundant
00:18:51 <ehird> & (for want of a better name) is simpler
00:19:12 <oerjan> you mean one taking only one character?
00:19:14 <ehird> all it does is: pushes the last command executed to the stack, quoted
00:19:17 <ehird> oerjan: no, simpler
00:19:20 <ehird> it's a regular command
00:19:27 <ehird> you just have one piece of global state that you assign with one line
00:19:42 <ehird> ais523 proved that this can do everything () can
00:20:09 <oerjan> well as long as you don't want to print non-command characters
00:20:26 <ehird> but let's remove output, it makes things easier to minimize; this is tarpitting after all
00:21:05 <ehird> & push last command executed, quoted
00:21:24 <oerjan> i think you might get rid of ! with some global rewriting in cps style, or something
00:21:26 <ehird> & is the most complex, but simpler than () in both implementation and syntax
00:21:29 <ehird> and about equal in semantics
00:21:36 <ehird> possibly simpler as you can leave non-command characters undefined
00:21:51 <ehird> hmm, let's make & , instead
00:21:56 <ehird> since it's so common, it might as well drop out
00:22:02 <ehird> or ', which is a quote
00:22:33 <ehird> so to push each command:
00:22:48 <ehird> dip being the hard part
00:22:55 <ehird> what's dip in underload again?
00:23:12 <ehird> oerjan: basically, in your interpreter loop, after the switch for the command, do lastcmd=thischar;
00:23:16 <ehird> so the answer is '
00:23:23 <ehird> if you do a fancy loop, you get the last thing executed there
00:23:39 <ehird> it's a variable declaration, one unconditional assignment and one trivial extra command
00:24:23 <ehird> oerjan: ok, first one done
00:24:30 <ehird> to push ~, ()()~'()!':*~a*^
00:24:47 <ehird> oerjan: there was one extra thing
00:24:51 <ehird> the stack starts out with () on it
00:25:00 <ehird> a better alternative is to say that the stack has an infinite number of ()s on the bottom
00:25:13 <ehird> that also handles things like the standalone program :
00:25:31 <ehird> ok then, to push ~, assuming initial stack:
00:26:10 <ehird> oerjan: of course you have to carry the () around with you, but iirc ais provided direct translations of every non-output instruction
00:26:16 <ehird> as in, simple text replacement
00:27:03 <ehird> happily, everything is now foo' on an empty stack which Just Works, even ''
00:27:15 <ehird> well, unless you need something before to be defined, which is reasonable
00:27:38 <ehird> anyway, I have a hunch that ! and * can be combined
00:28:23 <ehird> if we could rewrite dip without ~, we'd need no swap</captain obvs>
00:28:29 <oerjan> something that takes the top 3 on stack
00:28:38 <ehird> top 3 sounds inelegant
00:28:40 <oerjan> and combines two and drops one
00:28:56 <ehird> i was thinking more like,
00:29:24 <ehird> then you need a put-back instruction
00:29:33 <ehird> (foo)(bar)% = ((foo))(bar)^
00:29:39 <ehird> (foo)(bar)% = ((foo))bar
00:29:50 <ehird> this is how to drop
00:30:00 <ehird> you need to drop the extraneous ((foo)) after
00:30:17 <ehird> oerjan: is a required?
00:30:58 <ehird> (also adds unbalanced output!)
00:32:09 <ehird> but really, apart from ', which i think is a simplification, and dropping S, this is hard
00:33:24 <ehird> oerjan: maybe ^ could be simplified if we... holy fuck
00:33:30 <ehird> unify the data stack and the code.
00:33:36 * ehird 's mind, verily, blows
00:33:51 <ehird> that way we could remove ^.
00:34:01 <ehird> that is, if it's actually possible to
00:34:04 <oerjan> well to some degree ^ does nothing but move the top element of data to code
00:34:14 <ehird> we operate on code
00:34:41 <ehird> oerjan: heh, then ^ becomes the inverse of a
00:35:05 <ehird> (foo), stack/code is (separated by <>): (foo)
00:35:06 <oerjan> it already is, one way
00:35:16 <ehird> ^, stack/code is (separated by <>): foo<>^
00:35:28 <ehird> I guess ^ shouldn't be there
00:35:30 <ehird> ^, stack/code is (separated by <>): foo
00:36:16 <ehird> ((foo)^), stk/cd (sep <>): ((foo)^)<>^
00:36:41 <ehird> ^, stk/cd (sep <>): (foo)<>^
00:36:46 <ehird> (foo), stk/cd (sep <>): (foo)<>^
00:36:52 <ehird> ^, stk/cd (sep <>): foo
00:36:58 <ehird> oerjan: i think this could actually work
00:38:17 <ehird> oerjan: why? before we had two stacks, code and data
00:38:26 <ehird> this is simply removing a concept and that's it
00:38:31 <ehird> and it's wonderfully metacircular
00:39:26 <ehird> hey, i'm just minimizin' here
00:40:06 <oerjan> i'm not hatin', just lazyin'
00:40:44 <ehird> i'll try and run the traditional loop
00:41:07 <ehird> (:^), stk/cd (sep <>): (:^)<>:<>^
00:41:22 <ehird> we want to be executing the top item at all times
00:41:47 <ehird> initial stk/cd ^<>:<>(:^)
00:42:03 <ehird> the problem is avoiding an infinite loop there
00:42:18 <ehird> oerjan: (foo)'s semantics, I believe, are "move this (foo) down one in the stack"
00:42:34 <ehird> (foo)bar is bar<>(foo), and when executed it becomes (foo)<>bar
00:42:38 <ehird> and bar operates on the (foo) below it
00:43:17 <oerjan> you're aware that (:^):^ _is_ an infinite loop? ;)
00:44:10 <ehird> i just realised how (foo) should work
00:44:24 <ehird> initial stk/cd ^<>:<>(:^)
00:44:42 <ehird> (:^), new stk/cd: ^<>(:^)<>:
00:44:58 <ehird> :, new stk/cd: ^<>(:^)<>(:^)
00:45:09 <ehird> oerjan: i think every command has to move stuff out of the way, which is irritating
00:45:24 <ehird> i think the semantics are save second-top from us at beginning, pop it off; push it to top after we've done our stuff
00:47:21 <oerjan> logically the stack is equivalent to its elements being a'd and prepended to the code
00:48:19 <oerjan> and then you can just skip everything of (...) form to find the real next command
00:49:54 <oerjan> of course this messes with getting rid of ()
00:52:36 <ehird> I wonder if * shouldn't be what ~* is now
00:52:50 <ehird> also, skipping like that sucks; code should be executed from the top, as-is imo
00:53:25 <ehird> anyway, ~* seems damn common
00:53:33 <oerjan> however even with ~* as primitive i'm not sure you can do ~
00:53:33 <ehird> admittedly * has more elegance
00:54:28 <oerjan> because you still have no way to apply a to the bottom one before combining, if you don't have ~
00:55:23 <oerjan> face it, underload is _hard_ to improve
00:55:32 <oerjan> especially on elegance
00:55:36 <ehird> it is incredibly elegant.
00:55:56 <ehird> I wonder if we could reduce it to () + one command
00:56:07 <ehird> like a mammoth inelegant command looking at 9485345 values on the stack
00:56:47 <ehird> ~ and ! can easily be combined
00:56:52 <ehird> discard first element, swap rest
00:56:57 <ehird> then ! is ()combine
00:57:02 <ehird> swap rest two that is
00:57:16 <ehird> discard first element, concatenate second two, swap last two
00:57:20 <oerjan> um you mean ~ is ()combine
00:57:39 <ehird> ! is ()~()~combine
00:57:52 <ehird> then you have to pop afterwards :D
00:58:16 <ehird> oerjan: can we squeeze enclose second element in there?
00:58:58 <oerjan> the problem is when you want the combination to be two unenclosed things appended
00:59:32 <ehird> discard 1, enclose 2, cat 3&4, swap 5&6
01:00:10 <ehird> the question is how do you extract the commands from that...
01:00:23 <ehird> anyway with that it'd just be that+()+^+S
01:00:35 -!- augur has joined.
01:00:39 <ehird> discard 1, enclose 2, output 3, cat 4&5, swap 6&7
01:00:47 <ehird> (output also discards obvs)
01:00:58 <ehird> i'm not sure you can combine ^ into it
01:01:35 <ehird> oerjan: erm, 1 = TOS
01:01:40 <ehird> is that confusing?
01:02:04 <oerjan> i just shifted it so i didn't have to repeat the next
01:02:39 <ehird> also, cat 4&5 after output refer to elements after the output ofc
01:02:46 <ehird> i.e., when we drop we stitll use the same numbering
01:03:02 <ehird> run 0, discard 1, enclose 2, output 3, cat 4&5, swap 6&7
01:03:15 <ehird> first person to extract all the commands from that wins a cookie
01:03:33 <ehird> it is called either # or @, the most ... full ASCII characters.
01:06:14 <oerjan> hm... you're forgetting anything to do with :
01:06:53 <ehird> not worth thinking about until i know those commands can be extracted...
01:07:26 <oerjan> i expect there needs to be some tinkering about with the order of things to ensure you can extract
01:23:34 -!- coppro has joined.
01:33:03 -!- coppro has quit (Remote closed the connection).
01:33:57 -!- coppro has joined.
02:06:26 -!- ehird has quit (Read error: 113 (No route to host)).
02:15:12 -!- FireFly has quit ("Later").
02:39:08 -!- Pthing has quit (Remote closed the connection).
03:09:25 <Warrigal> Should I just subtract two from my age whenever I want to get his?
03:09:39 <Warrigal> Or perhaps I can say he's my little brother's age.
03:11:25 <oerjan> his birthday was a month or two ago iirc
03:13:13 <oerjan> i would hazard a guess that you can indeed subtract a rather accurate constant, unless space travel picks up considerably
03:47:33 <Warrigal> So, a bit younger than my little brother.
03:47:50 <Warrigal> Dear little brother: Please overachieve. Love, Warrigal.
04:33:52 -!- augur has quit (Read error: 104 (Connection reset by peer)).
04:34:04 -!- augur has joined.
05:12:34 -!- coppro has quit (Remote closed the connection).
05:15:56 -!- Pthing has joined.
05:25:38 -!- oerjan has quit ("leaving").
05:27:36 -!- coppro has joined.
05:57:27 -!- coppro has quit (Read error: 104 (Connection reset by peer)).
06:02:14 -!- coppro has joined.
06:06:08 -!- kar8nga has joined.
06:17:16 -!- pikhq has quit (Read error: 113 (No route to host)).
06:19:35 -!- pikhq has joined.
06:45:54 -!- kar8nga has quit (Remote closed the connection).
07:19:03 -!- CESSMASTER has quit (Read error: 104 (Connection reset by peer)).
07:19:20 -!- CESSMASTER has joined.
07:54:36 <fizzie> Sometimes the interweb speeds here at work make me a bit envious; I just fetched a 695 MiB Ubuntu image at an average speed of 28.1 MiB/s; that's about 236 Mbps. At home it's something like a fifteenth of that, and even that to one direction only.
07:56:45 <Deewiant> A fifteenth of that is still relatively fast
07:57:13 <fizzie> Yes, but a puny trickle compared to the datavalanche here.
07:59:59 -!- clog has quit (ended).
08:00:00 -!- clog has joined.
08:41:32 -!- MigoMipo has joined.
09:23:19 -!- Warrigal_ has joined.
09:23:19 -!- Warrigal has quit (Read error: 104 (Connection reset by peer)).
09:26:09 -!- Gracenotes has joined.
09:27:06 -!- FireFly has joined.
09:51:53 -!- augur has quit (Read error: 104 (Connection reset by peer)).
10:03:42 -!- Pthing has quit (Remote closed the connection).
10:26:16 -!- BeholdMyGlory has joined.
12:26:31 -!- BeholdMyGlory has quit (Remote closed the connection).
12:28:21 -!- coppro has quit (Read error: 110 (Connection timed out)).
13:42:36 -!- BeholdMyGlory has joined.
14:50:52 <AnMaster> fizzie, "datavalanche" ← very nice word
14:51:07 <AnMaster> fizzie, so how long did it take?
14:51:35 <AnMaster> fizzie, also, that would be gbit ethernet at work plus a very good uplink
14:52:04 <fizzie> I don't have the wget visible any more, but based on those numbers something like 25 seconds.
14:52:23 <AnMaster> lets assume you got theoretical max for calculating that... It means 28.1 MB/s is 224.4 Mbit/s ?
14:52:38 <AnMaster> which means quite an awesome fiber connection I bet
14:52:43 <AnMaster> fizzie, where is that you work!
14:52:48 <AnMaster> I want to work there too certainly
14:52:50 <fizzie> 28.1 MiB/s, so 236 Mbit/s.
14:53:20 <AnMaster> 28.1*8 == 224.8 says my calculator
14:53:29 <AnMaster> fizzie, also I always use base 1024 for MB and Mbit
14:53:38 <AnMaster> so I maintain that what I said above was correct
14:53:55 <fizzie> Telecomms people pretty much only use the non-binary mega in Mbps figures.
14:54:20 <fizzie> Anyway, it's still the same university.
14:54:48 <AnMaster> fizzie, anyway you probably got a bit less than theoretical max. (other users, network overhead, latency + tcp having to wait for confirmation and so on...) So maybe 250 Mbit/s?
14:55:30 <AnMaster> fizzie, what mirror was it btw?
14:56:02 <AnMaster> alternative explanation: University happens to host a mirror and you got a quite mediocre speed in those conditions
14:56:02 <fizzie> It was nic.funet.fi (well, fi.archive.ubuntu.com, but that's the same thing), so it's sort-of part of the same network, if you count the whole Funet -- http://www.csc.fi/english/institutions/funet -- as a one network.
14:56:17 <fizzie> It's not local to our university, but not too far away.
14:56:49 <fizzie> http://pastebin.com/m7a43072e -- not too many hops.
14:56:57 <AnMaster> fizzie, doesn't that give higher latency to the rest of the internet, I mean you have to pass through some central router from this network I guess
14:57:20 <AnMaster> fizzie, yeah that explains a bit of it
14:58:10 <fizzie> Probably, yes. Though it should be pretty fast to other nordunet university networks, there's a 10-gigabit uplink that way.
14:58:52 <Deewiant> FUNET is a member of NORDUnet.
14:59:26 <Deewiant> FUNET, SUNET, the Norway/Denmark equivalents, maybe something else
14:59:29 <fizzie> "Nordic countries"; I'm looking for the explicit list.
14:59:41 <Deewiant> Wikipedia has it: http://en.wikipedia.org/wiki/NORDUnet
15:00:12 <AnMaster> ah SUNET. But I don't remember my tracepath to uni (for some reason I can't even tracepath/traceroute a single hop away from UNI) went like tele2 -> telia for one or two jumps -> sunet
15:00:20 <fizzie> "Alcatel-Lucent and IPNETT has completed the 40 Gbit/s upgrade of the first link of NORDUnet's optical and IP transport network, which provides the Nordic infrastructure for research and education. Preparing the move to 100 Gbit/s, this network upgrade will further support new application enablement for universities and scientists in the Nordics and strengthen international collaborative environment." That sounds pretty fast.
15:00:36 <AnMaster> though today it seems to go straight from tele2 to sunet
15:00:43 <fizzie> Also buzzwordy. "Enablement", especially.
15:00:50 <fizzie> I feel very enablementy today.
15:01:06 <Deewiant> http://upload.wikimedia.org/wikipedia/commons/a/a3/FUNET_backbone.svg
15:01:25 -!- kar8nga has joined.
15:03:06 <AnMaster> Deewiant, those colours doesn't quite match the text...
15:03:22 <fizzie> 2009-09-21 17:03:12 (27.2 MB/s) - `tmp.exe' saved [108279664/108279664]
15:03:53 <AnMaster> the stuff going to denmark is violet kind of thing, and it says 20 gbit/s, but in the legend that colour is listed as 1 gbit/s or 622 mbit/s
15:04:11 <fizzie> That tmp.exe was a hundred-megabyte DirectX thing from Microsoft's download system; that should be further away, but still 27.2 MiB/s.
15:04:25 <Deewiant> It's a lighter purple, and as you say it says it's 20 Gbit/s so it's not messed up anyway
15:04:44 <AnMaster> Deewiant, lighter? I blame bad laptop^Wnotebook screen then
15:04:52 <AnMaster> even at other angles there is no difference
15:04:53 <fizzie> AnMaster: Went through Sweden, it seems.
15:05:28 <AnMaster> fizzie, well yes, it is so our FRA can log you :P
15:06:54 <fizzie> I think there was some talk years ago about Sonera relocating mail servers of Finnish customers away from Sweden because of that.
15:06:54 <fizzie> http://www.idg.se/2.1085/1.153711
15:06:54 <fizzie> Okay, a year and a bit ago.
15:08:12 <fizzie> According to FUNET_backbone.svg, the Internet is in Denmark.
15:08:26 * AnMaster laughs as a file about MSDNAA opens in open office
15:08:41 <AnMaster> oh and every other file at uni so far has been *.pdf not *.doc
15:10:23 <AnMaster> I'm trying to find the terms and conditions for it, but can't
15:10:53 <fizzie> Our "machine learning: basic principles" course assistant complained (at lunch) that people have been submitting just about everything, even though the instructions clearly say "single PDF file containing solutions to all questions"; there have been zipped up folders of separate files, Word documents, postscript files, postscript files renamed to .pdf, and even one PDF file named ".pdf" which was quite a clever trick, what with not showing up in ls listings.
15:13:16 -!- kar8nga has quit (Remote closed the connection).
15:13:39 <fizzie> There's a fancy Funet status graph at http://www.csc.fi/funet/status/tools/wm
15:15:44 <fizzie> Based on those line widths (although they are a bit unclear) it looks as if nowadays there are 10G uplinks from both Espoo and Helsinki to NORDUnet, and the link between those two is 10G too.
15:20:13 <AnMaster> "postscript files renamed to .pdf" <-- wonderful
15:20:43 <fizzie> Deewiant: The csc one, I think.
15:21:02 <fizzie> There's a similar graph of the NORDUnet backbone at http://stats.nordu.net/stat-q/load-map/ndn-map,,traffic,peak if you like looking at the rainbow-like colours.
15:21:04 <AnMaster> fizzie, what does the text in the pic mean
15:21:34 <AnMaster> hm and the colours are for percentages of what?
15:22:14 <fizzie> Percentages of maximum capacity for peak use during a day, or something like that.
15:23:11 <fizzie> "* = rented traffic capacity" and "other connections owned [by FUNET]: DWDM/CWDM/black fiber" are those labels in the lower-left corner.
15:24:19 <fizzie> Sorry, apparently "dark fiber" is the proper English term.
15:24:31 <AnMaster> hm can't see anywhere that it states you aren't allowed to use linux + wine + MSDNAA apps or such
15:24:33 <fizzie> "black" would've been the literal translation, though.
15:25:42 <fizzie> As long as you use it for study-related purposes; I think there was some sort of general condition for that.
15:26:25 <AnMaster> "Windows client operating systems obtained as part of the Developer AA program may not be used only as a terminal for accessing a UNIX telnet server"
15:27:26 <AnMaster> "The underlying Microsoft Windows desktop operating system may only be installed on new or replacement PCs acquired without an operating system for exclusive use by students or faculty enrolled in or employed by the member Department or organization; it is not transferable and must be removed if the PC is transferred."
15:27:52 <fizzie> Hmn. Well, I can't quite remember what terms there might've been. I just seem to recall something vague about the "only related to studies" thing.
15:30:02 <AnMaster> ok it says to activate I have to send a mail from my student mail to a specific internal address. Says nothing about subject or such.
15:30:26 <AnMaster> I'm going to ask about terms and conditions in the mail though. If it is auto-processed, it isn't my fault.
15:30:28 <fizzie> It wasn't called "developer AA" back then anyway, they might've rewritten parts of the terms.
15:31:08 <fizzie> I remember this one: "You may not use the software: -- to develop and maintain your own administrative or IT systems."
15:31:19 <fizzie> It's so awfully vague.
15:31:30 <AnMaster> fizzie, there are different variants, one that basically seems to be for arts students too
15:32:39 <fizzie> There's that virtualization thing explicitly, though: "You may also install such products in virtual machine environments so long as these virtual machines are (a) used according to these terms, and (b) run on machines owned or leased by you and/or students."
15:34:35 <fizzie> I guess the vagueness in "You may not use the software -- to develop and maintain your own administrative or IT systems" comes from the fact that I instinctively read that 'you' as, you know, myself, when in fact the "you" here is the university department.
15:38:11 <fizzie> (Gone for a while now.)
15:42:18 <AnMaster> well, for now I sent a mail to the IT support asking about what specific edition of the MSDNAA program is used here
15:44:01 <AnMaster> when apt-get update prints "Ign" or "Hit", what do those mean
15:44:23 <Deewiant> The first is probably "Ignore"
15:44:33 <AnMaster> Deewiant, yes but what do they mean in this context
15:45:11 <AnMaster> can't find anything in man page
15:45:27 <AnMaster> which updates the list of packages in the repos
15:46:08 <Deewiant> So when does it print that stuff then?
15:46:29 <AnMaster> Ign http://ddebs.ubuntu.com jaunty-security/universe Packages
15:46:29 <AnMaster> Ign http://ddebs.ubuntu.com jaunty-security/multiverse Packages
15:46:29 <AnMaster> Hit http://ddebs.ubuntu.com jaunty/main Packages
15:46:29 <AnMaster> Hit http://ddebs.ubuntu.com jaunty/restricted Packages
15:46:49 <AnMaster> those extra whitespaces at end, odd copy didn't strip that
15:48:05 <Deewiant> Does it print anything other than those?
15:49:00 <Deewiant> I mean, does it ever print anything other than Ign or Hit when you do an update
15:49:22 <AnMaster> when there is some change it prints also something like: Fetched 28.9kB in 3s (8912B/s)
15:49:28 <AnMaster> just before the "Reading package lists..."
15:49:38 <AnMaster> Deewiant, it prints a total of 91 Ign/Hit lines
15:50:04 <AnMaster> oh yes in case of new stuff it also prints:
15:50:06 <Deewiant> Then I guess Hit is just that it checked that package list
15:50:07 <AnMaster> Get:4 http://ddebs.ubuntu.com jaunty-security Release [14.2kB]
15:50:23 <AnMaster> Deewiant, so the hit line is still there
15:50:24 <Deewiant> Hit = checked package list, no changes
15:50:40 <Deewiant> Ign = didn't check package list
15:50:57 <AnMaster> Deewiant, well those extra repos it ignores are listed as "I want these"
15:51:45 <Deewiant> If you run apt-get update again, do you get all Ign?
15:52:04 <AnMaster> Deewiant, full output from http://pastebin.ca/1574011 from updating in a 32-bit chroot I use for cross compiling
15:52:14 <Deewiant> In that case Hit could've meant that something changed upstream but not necessarily anything interesting to you
15:52:32 <AnMaster> Deewiant, see there too, since nothing relevant changed, or apt-get upgrade would have said so
15:52:44 <AnMaster> like, it would have upgraded the packages
15:52:52 <AnMaster> also why would it say hit just half a minute later again
15:53:06 <Deewiant> AnMaster: Googling suggests that Ign may mean a 404
15:53:24 <AnMaster> Deewiant, ok that is strange, I tried google too
15:54:50 <AnMaster> Ign http://ddebs.ubuntu.com jaunty-updates/universe Packages
15:55:05 <AnMaster> http://ddebs.ubuntu.com/dists/jaunty-updates/universe/binary-i386/ exists and contains Packages
15:55:24 <Deewiant> I dunno, run with extra max verbosity or something
15:55:55 <AnMaster> -V Show verbose version numbers
15:56:13 <AnMaster> no change, guess it is just for package upgrades/installs
15:57:08 <AnMaster> can't find any "more verbose option" apart from that
15:58:19 <AnMaster> -q makes it loggable, so it doesn't use progress bars and such, and -qq makes it output errors only
16:00:22 <AnMaster> well there is -o for "set any configuration option" guess I will try to find a list of thosr
16:23:31 -!- Gracenotes has quit ("Leaving").
16:30:01 <AnMaster> do people actually write VLIW asm
16:30:12 <AnMaster> I assume they must, for stuff in kernel and such
16:56:22 -!- Asztal has joined.
16:56:52 -!- ais523 has joined.
17:02:30 * ais523 is trying to sort out university for this year
17:02:39 <ais523> apparently my course lasts a maximum of 4 years and a minimum of 6 years
17:02:51 <ais523> and until I pointed it out, nobody saw a problem with this
17:03:19 <AnMaster> ais523, does that mean "everyone thought that was normal" or "no one noticed the typo"?
17:03:37 <AnMaster> ais523, and what course is it that lasts that long
17:03:41 <ais523> I think it's that nobody had looked at both stats simultaneously
17:03:59 <ais523> and a 75% part-time PhD lasts 4 years (that's what I'm doing), if it had been 50% part-time it would last 6 years
17:04:07 <ais523> which is where the spurious 6-year figure seems to have come from
17:04:37 <AnMaster> ais523, what is the topic of the PhD?
17:04:54 <ais523> hardware synthesis from high-level languages
17:05:02 <ais523> well, higher level than VHDL, anyway
17:09:00 <AnMaster> that add/remove thing under the program menu in ubuntu sucks a bit:
17:09:21 <AnMaster> searching for RTS in the games section turns upp stuff like "Alien Blaster is a classic 2D shoot 'em up..."
17:09:32 -!- impomatic has joined.
17:09:33 <AnMaster> I can't even imagine how that matched RTS
17:09:52 <AnMaster> oh point and click adventures turn up too for that term
17:10:04 <AnMaster> and turned based stratergy, at least that is closer
17:10:32 <AnMaster> no actual RTSes though, not even the ones I know exist in the repos
17:12:01 <AnMaster> much better luck looking in /usr/portage/games-stratergy on gentoo, at least you only get turn based and real time ones, instead of getting completely unrelated arcade, board game,s side scrollers and what not
17:12:12 <AnMaster> gentoo has a lot fewer games though
17:13:54 <ais523> AnMaster: add/remove is a rather dumbed-down interface
17:13:59 <ais523> use apt-cache search instead
17:14:03 <ais523> the command-line stuff tends to work better
17:14:40 <AnMaster> ais523, not like that... I mean, at least gentoo has eix which has nice colour coded output
17:15:42 <AnMaster> while apt-cache search RTS lists lots of unrelated stuff too
17:16:00 <AnMaster> apt-cache search --help isn't useful either
17:16:26 <AnMaster> ais523, and this matches descriptions or tags I hope?
17:16:34 <AnMaster> because I'm not just looking for packages *named* RTS
17:17:07 <ais523> maybe games don't call themselves "RTS", but "real time strategy"?
17:17:17 <AnMaster> but stuff like "foobar is a 3D RTS which takes place in a post nuclear dystopian setting"
17:18:01 <AnMaster> that gives a few hits, but those I know about and I'm pretty sure it is missing some
17:19:00 <AnMaster> great, the website of the one new I found just give php errors, it looks interesting in google cache though
17:19:08 <AnMaster> Fatal error: Out of memory (allocated 2883584) (tried to allocate 311296 bytes) in /home/glob2/public_html/wiki/languages/messages/MessagesEn.php on line 811
17:20:31 <AnMaster> if it needs several MB to display a page
17:21:14 <AnMaster> ouch it wants even more boost fail libs
17:21:28 <AnMaster> libboost-thread1.34.1 this time
17:22:02 <AnMaster> and ubuntu package is as usual way outdated
17:22:24 <AnMaster> lifthrasiir, what was there to parse for going to a damn image page -_-
17:23:49 <ais523> lifthrasiir: what version of MediaWiki?
17:23:54 <ais523> they replaced the parser a couple of versions ago
17:24:00 <ais523> with one that isn't nearly as hackish
17:24:21 <lifthrasiir> ais523: i'm not sure, maybe older than 1.10?
17:24:34 <ais523> ok, that'll be the old parser then
17:25:06 <AnMaster> not exactly what I imagined, but doesn't seem bad as such
17:25:25 <lifthrasiir> anyway my server run several copies of mediawiki and they are becoming huge bottle neck... :S
17:25:57 <lifthrasiir> (to the point that ordinary bot can present practical DDoS attack to the wiki)
17:26:54 <ais523> lifthrasiir: even one copy of MediaWiki was enough to crash ehird's slice a while ago
17:32:14 <lifthrasiir> AnMaster: btw, esotope-bfc would become a topic of my bachelor's thesis. so the project is likely to be revived... soon.
17:32:32 <AnMaster> lifthrasiir, it was dead? And what on earth will your thesis be about?
17:32:56 <lifthrasiir> maybe about optimizing brainfuck compiler.
17:34:28 <lifthrasiir> and i don't know it was dead, but certainly i didn't work on it for months so... you know.
17:35:30 <lifthrasiir> (and of course the thesis would be in korean only. sadly i'm not good at academic writing in English)
17:40:09 <pikhq> Probably better than my academic writing in Japanese.
17:40:22 <pikhq> (my Japanese is about on par with that of a 2nd grader. Hooray?)
17:40:53 <lifthrasiir> of course i'm also not good at academic writing in Korean so that's not a problem. :p
17:58:33 <fizzie> AnMaster: If you want to find strategy-related games in the debian repos, just search for the game::strategy tag. (I'm not sure how to actually *do* that with apt, but the debtags smart-search can do it, at least.)
18:03:16 <AnMaster> lifthrasiir, in-between is dead
18:20:46 <fizzie> Yes, the correct aptitude /-search for that is "?tag(game::strategy)", or alternatively "~Ggame::strategy". I don't think Synaptic (the graphical apt GUI Ubuntu comes with, I think?) knows anything about searching for tags, though.
18:21:19 <fizzie> (And I guess there's some even more dumbed-down add/remove-apps thing?)
18:23:19 -!- adam_d has joined.
18:24:34 -!- ais523 has quit (Remote closed the connection).
18:41:20 -!- ehird has joined.
18:41:39 <ehird> 19:09:04 <Warrigal> Whenabouts was he born?
18:42:04 <ehird> 23:54:36 <fizzie> Sometimes the interweb speeds here at work make me a bit envious; I just fetched a 695 MiB Ubuntu image at an average speed of 28.1 MiB/s; that's about 236 Mbps. At home it's something like a fifteenth of that, and even that to one direction only.
18:42:10 <ehird> where do you work :|
18:42:39 <fizzie> It's there a bit later, but it's just that boring university.
18:44:52 <ehird> fizzie: better boring than shitty
18:45:01 <ehird> 06:56:02 <fizzie> It was nic.funet.fi (well, fi.archive.ubuntu.com, but that's the same thing), so it's sort-of part of the same network, if you count the whole Funet -- http://www.csc.fi/english/institutions/funet -- as a one network.
18:45:06 <ehird> oh you're on the same internet?
18:45:09 <ehird> that doesn't count then!
18:46:02 <fizzie> I can't be exactly sure where in Funet that Ubuntu mirror is.
18:46:12 <ehird> fizzie: i'm joking, i.e. funet is just a network like the internet
18:46:22 <ehird> so, hyperbolically, if funet is discounted the internet is!
18:47:22 <fizzie> Yes, you can only count downloads from the moon or Mars.
18:47:43 -!- impomatic has quit (Read error: 110 (Connection timed out)).
18:48:07 <ehird> if I double click a word to select it in ubuntu
18:48:10 <ehird> then drag from there
18:48:14 <ehird> it makes a selection from my cursor position
18:48:17 <ehird> e.g. the middle of the word
18:54:12 -!- sebbu2 has joined.
19:00:18 -!- sebbu3 has joined.
19:03:46 <ehird> 07:30:26 <AnMaster> I'm going to ask about terms and conditions in the mail though. If it is auto-processed, it isn't my fault.
19:03:48 <ehird> That's, uhh, false.
19:06:33 <ehird> AnMaster: Hit = Going there, nothing
19:06:40 <ehird> Ign = I don't want to look at this
19:06:50 <ehird> Get = Lol I be outdated, like de dinosorz
19:07:18 <ehird> 07:52:04 <AnMaster> Deewiant, full output from http://pastebin.ca/1574011 from updating in a 32-bit chroot I use for cross compiling
19:07:24 <ehird> Gee, I'd try and help but you set that to expire.
19:07:49 <Deewiant> It just had two Ign and two Hit lines
19:09:02 <Deewiant> Or no, that was the full output
19:09:28 -!- sebbu has quit (Connection timed out).
19:09:29 -!- sebbu3 has changed nick to sebbu.
19:10:16 <ehird> 08:30:01 <AnMaster> do people actually write VLIW asm
19:10:17 <ehird> 08:30:08 <AnMaster> if so, what does it look like
19:10:17 <ehird> 08:30:12 <AnMaster> I assume they must, for stuff in kernel and such
19:10:20 <ehird> See Itanium documentation.
19:11:00 <ehird> 09:03:37 <AnMaster> ais523, and what course is it that lasts that long
19:11:07 <ehird> Does Sweden not have PhDs or something
19:11:45 <ehird> 09:10:44 <AnMaster> complete and utter fail
19:11:57 <ehird> Waah, the search engine doesn't magically expand my algorithm and search with a strong AI
19:12:07 <ehird> I'm going to act like a 7 year old and say fail
19:12:19 <ehird> 09:13:54 <ais523> AnMaster: add/remove is a rather dumbed-down interface
19:13:58 <AnMaster> <ehird> It looks boring. <-- meh
19:14:24 <ehird> http://sverre.web.cern.ch/sverre/IA64_1.pdf
19:14:27 <ehird> Tutorial with examples
19:14:42 -!- sebbu2 has quit (Connection timed out).
19:15:58 <ehird> 09:14:40 <AnMaster> ais523, not like that... I mean, at least gentoo has eix which has nice colour coded output
19:15:59 <ehird> 09:14:44 <AnMaster> and nicely formatted
19:16:04 <ehird> You know, you could stop complaining and use Gentoo.
19:16:20 <ehird> 09:17:17 <AnMaster> but stuff like "foobar is a 3D RTS which takes place in a post nuclear dystopian setting"
19:16:30 <ehird> Debian usually opt for legible descriptions as opposed to requiring niche knowledge.
19:17:01 <ehird> 09:21:14 <AnMaster> ouch it wants even more boost fail libs
19:17:05 <ehird> Dependencies! Astonishing!
19:17:42 <ehird> 09:32:56 <lifthrasiir> maybe about optimizing brainfuck compiler.
19:17:50 <ehird> please, please, please use the proper name of the language
19:17:59 <ehird> everything i've seen that tries to be serious about it says BF :(
19:18:27 <ehird> "On Advanced Optimization Techniques for the Abstract Imperative Language 'Brainfuck'"
19:18:37 -!- lament has joined.
19:18:42 <FireFly> :(, counter-increment: sum attr(id); no workie
19:18:43 <lament> map map map map fold fold fold fold
19:18:54 <ehird> "I don't think Synaptic (the graphical apt GUI Ubuntu comes with, I think?)"
19:18:59 <lament> select select select select aggregate aggregate aggregate aggregate
19:19:05 <ehird> The package manager, yes. Not the more friendly "Add/Remove..."
19:19:14 <ehird> lament: increment increment decrement decrement
19:20:27 <ehird> You know, the default radeon/ati driver kinda sucks when paired with Adobe Flash.
19:20:43 <ehird> As in "music is desynchronized by like a 1/4 second" sucks.
19:20:49 <ehird> Well, not just music.
19:20:53 <ehird> Like, every piece of audio.
19:21:01 <ehird> It doesn't look for music and then delay just that.
19:22:02 <fizzie> The delay is a function of the musicalnessity of the sound; it wouldn't delay, say, rap music, much if at all.
19:24:38 <fizzie> It's funny how Spotify under Wine keeps saying "fixme:ntdll:NtQueryInformationProcess (process=0xffffffff) Unimplemented information class: ProcessDebugFlags" once per second. I wonder if it's doing some sort of anti-anti-DRM "explode if someone's trying to use a debugger to extract song data" check, or if that's just unrelated Wine coincidence.
19:24:53 -!- kar8nga has joined.
19:24:58 <ehird> Torrents don't do that :-P
19:25:33 <fizzie> But it's not as user-friendly to browse around with torrents. Besides, I wouldn't get the wonderful Spotify ads then.
19:26:06 <ehird> Someone ought to make a content-type tailored torrent searcher.
19:26:36 <ehird> Although torrents will always be quite slow, even on a fast connection.
19:31:20 <fizzie> I have this Sennheiser USB microphone/speaker headset -- well, it has the normal 3.5 mm headphone/mic plugs, and an USB -> 2x3.5mm adapter -- and when I plug it in, it is detected as a HID device.
19:31:26 <fizzie> input: Sennheiser Communications Sennheiser USB Headset as /class/input/input5
19:31:26 <fizzie> generic-usb 0003:1395:0002.0004: input,hidraw3: USB HID v1.00 Device [Sennheiser Communications Sennheiser USB Headset] on usb-0000:00:12.2-1.2/input3
19:31:58 <ehird> Move your mouse by talking into it.
19:32:12 <ehird> "Up. Up. Left. Right. No, a bit to the left. Click."
19:33:26 <fizzie> There's an Audio/Control Device interface, a couple of Audio/Streaming interfaces, and then a "No Subclass" HID device.
19:36:09 <fizzie> Heh, things are nowadays so complicated, with every device having all kinds of crazy parts. According to ALSA I have four sound cards, for example; one's the motherboard-integrated, one's this USB thing, one's the TV capture card, and one's the Radeon graphics card HDMI port.
19:43:05 <ehird> fizzie: The Radeon card actually works as a sound card, I think.
19:43:17 <ehird> Who knows how it compares to onboard.
19:44:03 <fizzie> I don't think I can get anything else than the digital s/pdif out of it, though. (And actually it's the onboard-integrated radeon, not a separate card. :p )
19:45:16 <fizzie> Anyway, the point of the exercise was to test whether with this USB adapter I could connect the better Sennheiser headphones to the computer without hearing the awful "you can hear the mouse moving" noise the chassis front panel headphone connector gives. (Yes.)
19:46:00 <ehird> Hearing the mouse moving sounds awesome.
19:46:53 <fizzie> I'm not sure if it's the USB-caused interrupts or what; you can also hear things happening on the screen; well, either that, or just the related computation.
19:47:27 -!- oerjan has joined.
19:50:24 <fizzie> No sign of noise with this adapter, and since it's Sennheiser(TM), I'm sure it can't be completely awful w.r.t. audio quality in general, though I'm not really qualified to judge that.
19:51:26 <ehird> fizzie: I thought that too, until I heard sound emanating from a wireless headphone of theirs.
19:51:36 <ehird> Then again it was more the whole radio thing and less the actual headphones that did that.
19:52:15 <fizzie> Ooh, I had radio-based no-name-brand wireless headphones (because the cat kept eating the wires off the wired ones whenever I left them around), and I've seldom heard a worse thing.
19:54:22 <oerjan> 09:02:39 <ais523> apparently my course lasts a maximum of 4 years and a minimum of 6 years
19:54:25 <oerjan> 09:02:51 <ais523> and until I pointed it out, nobody saw a problem with this
19:54:45 <ehird> you get the phd accepted, submit it, write it, study the advanced stuff
19:54:49 <ehird> and it gradually gets more basic
19:56:03 <fizzie> Official length for PhD studies around here is 4 years, with the theoretical studies done during the first two years, and the dissertation and related research mostly during the other two.
19:56:37 <fizzie> Should go and fish around for some signatures, my graduate-studies application form needs to be returned by Friday.
20:13:44 -!- Azstal has joined.
20:27:40 -!- Asztal has quit (Read error: 110 (Connection timed out)).
21:13:33 -!- ais523 has joined.
21:14:25 -!- ehird has quit (Remote closed the connection).
21:14:56 <ais523> hmm.. I can't even say hi back again
21:14:58 <ais523> because you aren't here...
21:16:22 -!- ehird has joined.
21:16:39 <ehird> I'd be fine if this radeon driver was just slow, but I'm pretty sure it's making Compiz unstable
21:17:56 -!- MigoMipo has quit ("Page closed").
21:18:07 <ais523> you left before I could say hi
21:18:14 <ehird> that's what she said?
21:19:22 * ehird makes metacity compositing with gconf-editor because compiz is just so unstable with this driver
21:19:33 <ehird> and I need my damn window shadows
21:20:07 <ais523> I didn't realise metacity could composit
21:20:07 <ehird> unfortunately compositing metacity's window shadows have a bit too much radius and blur
21:20:22 -!- oerjan has quit ("Later").
21:20:37 <ehird> /apps/metacity/general/compositing_manager
21:20:43 <ehird> you don't even need to restart it
21:20:56 <ehird> oh, and it draws the panel's window shadow underneath the windows
21:21:16 <ehird> which means you can put windows at the top or bottom and they won't look awful when unfocused
21:21:35 <ehird> (the shadow interacts with the beige inactive titlebar rather unpleasantly)
21:21:41 <ehird> apparently the compiz folks are trying to fix that though
21:22:56 <ehird> also, Ubuntu users: System -> Administration -> Software Sources -> Download from: -> Other... -> Select Best Server
21:23:09 <ehird> now apt-get update finishes in 0.6 seconds for me
21:23:15 <ais523> and recommend it over in #ubuntu whenever a new release comes out
21:23:15 <ehird> so everything is wonderfully fast
21:23:46 <ais523> I use the official Swedish server
21:23:46 <ehird> ais523: I got the same server trying it again about 5 minutes later, so this one I know is the best
21:23:54 <ehird> Oddly, it isn't in the UK
21:24:18 <ehird> nl2.archive.ubuntu.com
21:24:28 <ehird> ais523: why does it ping, though? that's not really an accurate test of total download speed, is it?
21:24:37 <ais523> and I do my scans just after release, which is when the Ubuntu servers are likely most overloaded
21:24:42 <ehird> I'd be willing to run something that takes 10 minutes instead of 2 to get a more accurate measure
21:24:50 <ais523> and although pinging isn't a good test of throughput, it /is/ a good test of congestion
21:25:11 <ais523> which is likely to be more important in general, as a congested server will take hours, whereas all non-congested servers will have similar times
21:25:15 <ehird> true, it's nice to have the Reload stuff in Add/Remove... and the like be instant
21:25:52 <ehird> (I'm quite pedantic about naming it with the ellipsis because Gnome has proper conventions for these things and I want them to stick)
21:26:09 <ehird> I should say GNOME, really, but it's not an acronym and that's an awfully shouty name
21:26:22 <ehird> not exactly humane :-P
21:26:42 <ehird> ais523: you use Epiphany sometimes, right?
21:26:48 <ais523> I have Epiphany open atm
21:27:08 <ais523> links from IRC open in Epiphany for me rather than Firefox
21:27:16 <ehird> ais523: a few questions that you may know - one, why did they remove epiphany-webkit in Jaunty so you have to add two webkit-dev PPAs and a PGP key?
21:27:26 <fizzie> Awesome has a compositing manager thing too; but literally the only thing it does is it can apply opacity values to active/inactive windows.
21:27:26 <ehird> also, you can configure that.
21:27:46 <ehird> btw, xmag makes quite a nice colour picker
21:27:50 <ais523> ehird: I use epiphany-mozilla atm
21:28:01 <ehird> you can select the general region then hover over individual pixels to find their rgb
21:28:14 <ehird> as (foo,bar,baz) where each element is repeated once
21:28:15 <ais523> and my guess is they removed it because nobody was willing to maintain it
21:28:18 <ehird> due to 16-bitness or something
21:28:30 <ehird> (f8f8,eaea,9797) etc
21:28:39 <ehird> ais523: wtf is epiphany-mozilla?
21:28:40 <ais523> ehird: that repetition would imply a 24-bit colour depth
21:28:52 <ais523> err... I mean epiphany-gecko
21:28:54 <ehird> right, which is standard
21:28:57 <ais523> it's been a long day, OK
21:29:08 <ehird> ais523: epiphany-gecko is all Ubuntu has right now, but Epiphany are going Gecko-only in the future
21:29:33 <ais523> I thought they were going Webkit-only
21:29:44 <ehird> btw, do you know where I can get a non-bitmap Chinese/Japanese font? annoying seeing all those pixelated characters among smoothed ones
21:29:52 <ais523> anyway, Epiphany does fit into Gnome really well
21:30:00 <ehird> yes, which is why I'd rather use it instead of Firefox
21:30:07 <ais523> even if it's not ideal as a browser itself, it's good enough, and the rest of the system really knows what it's doing
21:30:37 <ais523> my major annoyance with Epiphany atm is that you can't open a new tab by double-clicking the tab bar
21:30:40 <ehird> My favourite browser is Chrome, but it still draws that awful blue title/tab bar it uses on Windows XP and non-Aero Vista/7 in the super-alpha Linux version
21:30:49 <ehird> which makes it unusably puke-esque
21:30:50 <ais523> I should report that, it's not like double-clicking the bar does anything useful atm
21:31:07 <ehird> ais523: I had another complaint with the tab bar, but
21:31:11 <ehird> report it to the Gnome people
21:31:18 <ehird> they'll just use the stock Gnome tab bar, obviously
21:31:27 <ehird> or at least follow the Gnome conventions for tab bars strictly
21:31:35 <ehird> ais523: you can't middle-click to close a tab
21:31:39 <ehird> you have to hit the close button
21:31:46 <ehird> it's the same in Safari, but fucking annoying
21:32:00 <ais523> you're right, you can't
21:32:05 <ehird> I was using the PPA WebKit version, so probably not the same version as you (more recent)
21:32:11 <ehird> it has an uglier icon to the repository versions.
21:32:21 <ehird> yep, the Nautilus tab bar is identical
21:32:24 <ais523> yep, middle-clicking doesn't close tabs in Nautilus either
21:32:39 <ehird> anyway, I do that literally all the time, so it's very annoying
21:32:42 <ais523> at least it's not like Konqueror, where middle-clicking tries to interpret the selection contents as an URL and navigates there
21:32:57 <ehird> also if you middle click it goes to the URL in your clipboard
21:32:59 <ais523> it also causes you to lose the contents of the current tab as a result
21:33:10 <ais523> ehird: "also"? that's what I said
21:33:10 <ehird> Konqueror is insanity-inducing
21:33:17 <ehird> ais523: "selection contents"
21:33:22 <ehird> oh, is that the name of the X11 clipboard?
21:33:29 <fizzie> I think my Kanji characters and friends come from the ttf-{kochi,sazanami}-* packages, which I think are vector fonts, but I can't really say anything about the quality, since I can't decode the characters.
21:33:33 <ehird> huh, if you deselect it does indeed uncopy it
21:33:34 <ais523> I think it's technically called "primary selection" or something like that
21:33:58 <ehird> btw, Ubuntu's font rendering is quite passable in 9.04; they're using a new subpixel hinting thing
21:34:08 <ais523> I know, because I spent far too long fiddling with elisp trying to get copy-and-paste working between Windows and Emacs running on SunOS inside Exceed
21:34:23 <ehird> I'm using Slight hinting + subpixel right now, and I very rarely see any coloured fringes
21:34:26 <ais523> I managed it in the end, though; it was the easiest way to transmit information between the two systems
21:34:48 <ehird> ais523: anyway, another Epiphany query
21:34:54 <ehird> you know those Smart Bookmarks you drag to the toolbar?
21:34:56 <fizzie> There's a bit of add-on confusion in that X11 has both the primary selection and a separate "clipboard" selection.
21:34:57 <ais523> also, I never see coloured fringes on here with any settings but ones with the pixel order wrong
21:35:00 <ehird> or the toolbar with the address bar in
21:35:17 <ehird> ais523: darn. see http://projects.gnome.org/epiphany/smartbookmarks.html
21:35:28 <ais523> I don't have bookmarks on the toolbar at all in Epiphany
21:35:32 <ais523> instead I have them on the address bar
21:35:38 <ehird> basically, if you have a bookmark like the "Search the web" one included, with %s in it
21:35:42 <ehird> ais523: the address bar is on a toolbar
21:35:47 <ehird> anyway, you can drag it to any toolbar
21:35:52 <ehird> and it'll appear with a text field to the right
21:35:57 <ehird> ais523: right, I know that
21:35:59 <ais523> except programmed by hand rather than automatic
21:36:09 <ehird> that's useless, as you have to wait for a second for the completions to come up
21:36:11 <ehird> then hit down and enter
21:36:15 <fizzie> And I think there were something called "cut buffers" too?
21:36:16 <ehird> so it's worthless for googling
21:36:19 <fizzie> http://www.jwz.org/doc/x-cut-and-paste.html
21:36:29 <ehird> ais523: anyway, try dragging "Search the web" next to the address bar and you'll see
21:36:53 <ehird> the problems are twofold: one, you can't make it bigger, so the text box is always irritatingly small
21:37:00 <ehird> two, you can't assign the text box a keyboard shortcut
21:37:11 <ehird> so I can't do Ctrl+K hello Enter to google for something like I do now
21:37:37 <ehird> in conclusion, I can't find a way to search the web sanely with Epiphany
21:37:48 <ais523> you could always just visit Google and use the text box there
21:37:53 <ais523> that's not as good as the Firefox method, though
21:38:00 <ais523> it's not a problem I'd noticed, though
21:38:00 <ehird> ais523: do you have *any idea* how much I google?
21:38:04 <ais523> probably because I hardly use Google
21:38:10 -!- BeholdMyGlory has quit (Remote closed the connection).
21:38:24 <ehird> 1. type in query, wait a small amount of time, hit down to "Search the web" smart bookmark, enter
21:38:43 <ehird> 2. click dragged-to-toolbar "Search the web"'s input box, type query, enter (fastest out of these three)
21:38:45 <fizzie> If you want a point of reference, I just google with Ctrl-L "g hello" enter.
21:39:01 <ehird> 3. type in google.com, enter, wait for it to load, type query, enter
21:39:02 <ais523> oh, fun fact: today I was using someone else's computer (windows XP + IE), and I entered a URL wrong (it was one of the insane ones that need www. and don't work without)
21:39:12 <ehird> ais523: out of these, three is obviously the slowest by far
21:39:17 <ais523> so it decided to interpret my entry as a Bing search rather than going to the website
21:39:31 <ais523> Bing found the site as a result (nonsuprisingly, as I'd given it all the URL apart from the www)
21:39:32 <fizzie> Can't you do a smart-bookmark-with-a-"g"-keyword like that in Epiphany too?
21:39:36 <ais523> and when I clicked on the link, I got a security error
21:39:46 <ais523> on the basis that an untrusted website (Bing) was trying to link to a trusted one
21:39:59 <ehird> fizzie: I don't think so
21:40:06 <ehird> fizzie: even if you can, that's suboptimal
21:40:14 <ais523> this didn't really fill me with confidence in IE's usability...
21:40:25 <ehird> IMO, smart bookmarks are really stupid; the only use for them can be replaced with something like Firefox's search bar
21:40:39 <ehird> easily accessible, and you can easily add new engines with the OpenSearch thing that lets you add one from a compliant site
21:40:40 <ais523> they should just implement opensearch
21:40:46 <fizzie> There's no waiting involved, which (to me) is enough; because I really don't like the search bar.
21:40:55 <ehird> in conclusion, the Epiphany people are either just trying to be different for "simplicity" and failing wildly,
21:41:16 <ais523> my conclusions about Bing: it's not incredibly bad, but it still has no reason for existing
21:41:24 <fizzie> Because it looks so useless, there's already one text box up there, and it's bigger too.
21:41:26 -!- Asztal has joined.
21:41:37 <ehird> fizzie: Ctrl-K, Ctrl-L g space
21:41:44 <ais523> it doesn't seem to have an obvious way of making money either
21:41:51 <ehird> also, do you have a really small screen or something?
21:41:58 <ais523> perhaps it has adverts that I mentally filtered out
21:42:07 <ais523> that version of IE certainly wasn't running adblock
21:42:14 <ais523> maybe it had adverts that I didn't notice...
21:42:20 <ehird> http://www.bing.com/search?q=ads&go=&form=QBLH&filt=all&qs=n
21:42:27 <ehird> taking up the whole third column
21:42:34 <fizzie> ehird: Still no waiting, and (again, for me) the g-space prefix is just a small-constant O(1) thing when compared to writing the search query itself.
21:42:41 <ais523> (nowadays, I'm using adblock to block avatar folders on forums, so I don't have to look at everyone's awful avatars)
21:42:54 <ehird> fizzie: yes, but if you can remove the constant without downside, why not?
21:42:59 <ais523> ehird: ah good point, it does have adverts
21:43:03 <ais523> I had to conciously look at them to see them, though
21:43:17 <ehird> that's the same with most people.
21:43:20 <ais523> is there something wrong with me?
21:43:22 <fizzie> It's not "without downside", it eats up space from the address bar; and the search-bar itself is so short you can hardly fit any longer queries there.
21:43:44 <ehird> fizzie: for one, how tiny are your browser windows, and how long are the URLs of the sites you use?!
21:43:53 <ehird> for two, text boxes can accept more than their width, you know
21:43:59 <ehird> and my search box isn't *that* small
21:44:01 <fizzie> Er, well, admittedly my main browser window is 1920 pixels wide nowadays.
21:44:16 <fizzie> I don't know, it's probably really just that I never got used to it, before getting the habit of the C-l g-googling.
21:44:59 <fizzie> And on the 1024x768 iBook the address bar (with the toolbar buttons to the left, and a reasonably-sized search bar to the right) *was* a bit short-ish.
21:45:41 * ais523 wonders what the smallest usable text box is
21:45:47 <ais523> 4 characters is probably enough to correct typos
21:46:32 <ehird> usable as in barely or usable as in usable?
21:47:43 <ehird> "keyboard shortcut to smart bookmark epiphany" on google, first result: "Epiphany is hype, get over it"
21:47:51 <ehird> which is a positive post about Epiphany...
21:47:58 <ehird> http://raphael.slinckx.net/blog/2006-03-15/epiphany-is-hype-get-over-it
21:49:04 <ais523> incidentally, what's the appropriate reaction when you go to the official website for a TV show, and find that it's on Geocities and most of the links go to under-construction pages?
21:49:22 <ehird> anyway, in an unsurprising move a new entry on my todo list has been created, that of "make a really *Gnome* IRC client featuring integration with Ubuntu's fancy messaging menu thing"
21:49:46 <ehird> and since you don't know what that is unless you use Pidgin or... one other email client, probably,
21:50:07 <ehird> https://wiki.ubuntu.com/MessagingMenu
21:50:16 <ehird> well, I think Empathy integrates it too actually
21:50:27 <ehird> but I certainly went WTF when I saw it the first time...
21:50:32 <ehird> ais523: how old is this show?
21:51:09 <ais523> ehird: it's apparently very popular in France, but ran for a few years over in the UK and then went to satellite
21:51:21 <ehird> does it look Geocities?
21:51:29 <ais523> it even has that advert sidebar thing
21:51:33 <ehird> stop watching it :P
21:51:36 <ehird> ais523: all geocities pages have that...
21:51:48 <ais523> I got pretty good at Geocities back when it was popular
21:51:53 <ais523> ehird: you could have banner ads instead
21:51:56 <ehird> well, it's mandatory nowadays at least
21:52:00 <ais523> which weren't JS-laden, and so tended to work better than the sidebar
21:52:03 <ehird> ais523: I never saw that when I used it...
21:52:09 <ehird> I mean, even on geocities sites
21:52:16 <ehird> also, I used to use a script to hide the bar
21:52:20 <ais523> anyway, I used to connect a geocities site to a free domain name person
21:52:25 <ais523> in such a way that the ads cancelled each other out
21:52:35 <ais523> due to some sort of javascript crash
21:52:45 <ais523> I have no idea if it still exists
21:52:45 <ehird> those are the ones I used...
21:53:03 <ehird> hey, konversation was ported to kde 4
21:53:09 -!- Azstal has quit (Connection timed out).
21:53:14 <ehird> it's, uh, still ugly
21:53:28 <ais523> <Help|About> Version 4.2.2 (KDE 4.2.2)
21:53:36 <ehird> http://konversation.kde.org/screenshots/konversation12-alpha5_3.png ;; heh, this is from an old version... those quick buttons beneath the user list are from X-Chat
21:53:44 <ehird> and are perhaps the ugliest interface ever devised
21:53:58 <ehird> i mean, wow, do you really deop people that much?
21:54:00 <ais523> Konversation doesn't fill me massively with confidence in terms of programming quality
21:54:18 <ais523> although they were very responsive when I sent them a bug report
21:54:24 <ais523> it was a great bug, btw
21:54:30 <ehird> Konversation was my preferred IRC client in 2006 using Kubuntu
21:54:35 <ais523> typing /msg then a channel opened a second tab for the channel
21:54:40 <ais523> which could send but not receive
21:54:45 <ehird> the version before Kubuntu turned to purple, iirc
21:54:50 <ehird> ais523: Remember when Kubuntu turned purple?
21:54:55 <ehird> that release was so ugly...
21:55:11 <ais523> I've given up on trying to track irrelevant Konversation UI changes
21:55:18 <ehird> http://pillateunlinux.files.wordpress.com/2008/11/4-kubuntu610-1.png
21:55:19 <ais523> and I don't think I used KDE that release
21:55:37 <ehird> window borders: http://openlife.cc/system/files/snapshot1-640x400.png
21:55:46 <ais523> that link isn't too bad apart from the cog things in the menu sidebar
21:55:55 <ehird> the first thing I did that release was change the theme, or at least its colours
21:55:58 <ehird> ais523: the purple, man!
21:56:02 <ehird> it was overpowering
21:56:05 <ais523> I don't consider that colour to be purple
21:56:15 <ais523> (but then, colours are perceived differently by different people)
21:56:17 <ehird> it's the ugliest shade of blue, then
21:56:28 <ehird> ais523: note that that screenshot is darker than it was
21:56:30 <ehird> http://openlife.cc/system/files/snapshot1-640x400.png
21:56:35 <ehird> see the sidebar background
21:56:42 <ehird> that's the colour it was
21:57:26 <ehird> but, I mean, the release before was http://www.oracle.com/technology/tech/linux/install/files/anomalieaftereboot.png (except that screenshot's tinted too)
21:57:30 <ehird> so it was a lot... bluer
21:57:46 <ais523> ouch, that previous release is far too bluish-cyan
21:57:57 <ais523> I'd have to darken the background of that one to avoid hurting my eyes
21:58:11 <ehird> ais523: WTF? that's not cyan
21:58:19 <ais523> it's cyaner than pure blue is
21:58:20 <ehird> are you sure your display is adjusted right?
21:58:25 <ehird> it's very much blue
21:58:36 <ais523> try changing the alpha
21:58:39 <ais523> it obviously goes cyan when lightened
21:58:46 <ais523> whereas, blue goes lilacy when lightened
21:58:49 <ehird> ais523: no, my display is calibrated...
21:59:14 <ehird> that is totally blue, man.
21:59:18 <ais523> how do you extract RGB codes from xmag?
21:59:26 <ehird> hover over a pixel
22:00:05 <ais523> that's substantially greener than it is red
22:00:13 <ais523> (/me removed the repeats)
22:00:17 <ehird> it's slightly purplish blue...
22:00:30 <ais523> it's the oracle link that gave that colour
22:00:43 <ehird> rtttttttttfffffffffffffffffffffffffffffffffffffff444444444444444444444455555555555555555554555555555555555555555555555555oh crpa
22:01:08 <ehird> well this is annoying
22:01:13 <ais523> the openlife.cc title bars are 6977a6
22:01:27 <ehird> ais523: openlife.cc is 6.10
22:01:30 <ehird> my last link was 6.04
22:01:32 <ais523> again very slightly greener than red, although that's almost exactly pure blue
22:01:35 <ehird> which is the non-puple
22:01:35 <ais523> yes, I'm comparing the two links
22:01:49 <ais523> the one you say is non-purple I say is bluish-cyan, and the hex agrees
22:01:49 <ehird> title bas a ielevant
22:02:04 <ehird> i'm really not interested in this
22:02:05 <ais523> the one you say is purple I say is blue, and the hex agrees
22:02:09 <ais523> but yes, this is a silly topic
22:02:13 <ehird> ais523: no, look at the sidebar
22:02:20 <ehird> the Appearance selection, say
22:02:49 <ehird> EdgyEft/RC/Kubuntu - Ubuntu Wiki
22:02:50 <ehird> For information on previous releases of Kubuntu 6.10, the Edgy Eft, ... it has been decorated with the latest purple scheme, and the Kubuntu gears logo. ...
22:02:58 <ais523> greener than it is red
22:03:00 <ehird> [Bug 62618] edgy default purple color ugly
22:03:00 <ehird> 27 Sep 2006 ... [Bug 62618] edgy default purple color ugly. Marc Schiffbauer ... theme (window decorations and background) that kubuntu 6.10 has as default. ...
22:03:00 <ehird> www.mail-archive.com/ubuntu-bugs@lists.../msg60562.html - Cached - Similar -
22:03:03 <ehird> Artwork/Incoming/Attic/Kubuntu-Edgy-Ideas - Ubuntu Wiki
22:03:03 <ehird> 6 Aug 2008 ... The vision I have for Kubuntu Edgy is defined by use of color (purple and blue), texture (bubbles and/or a real pic and/or a mix of the two) ...
22:03:06 <ehird> Kubuntu Edgy Eft (Glass) KDE-Look.org
22:03:06 <ehird> Kubuntu Edgy Eft (Glass) KDE-Look.org Community Portal for KDE Themes ... of) and purple for the purists and traditionalists who'll yell "Edgy is purple! ...
22:03:14 <ehird> ais523: Kubuntu 6.10 was purple, and I rest my case.
22:03:15 <ais523> clearly, the names I use for colours match up with the ones that computers use, but not the ones the general public use
22:03:20 <AnMaster> <ehird> 09:03:37 <AnMaster> ais523, and what course is it that lasts that long
22:03:20 <AnMaster> <ehird> Does Sweden not have PhDs or something
22:03:26 <ehird> ais523: including the people behind it
22:03:50 <ais523> I'm just surprised that people can call 7579ba purple
22:03:58 <ais523> but then, most people consider, say, ddddff purple
22:04:01 <AnMaster> do you use "course" both for the full thing and for stuff like "networking in java"?
22:04:21 <ais523> although it's only correct usage for the full thing
22:04:24 <ais523> the second would technically be "module"
22:04:28 <ais523> but people call it "course" often anyway
22:04:55 <ehird> incidentally, I find it odd that Gnome has a bunch of disparate download folder settings, and each app does its own download manager
22:05:21 <Deewiant> ais523: What should one call 7579ba
22:05:23 <ehird> the Gnome way would be to have a folder ~/.gnome2/downloads that every app downloads to, and when you open it in Nautilus it's special (like the Wastebasket or a CD)
22:05:31 <ais523> Deewiant: I'd consider it a shade of blue
22:05:33 <ehird> it always displays as a list, and you have progess bars/a cancel button
22:05:38 <ehird> plus that orange header saying "Downloads"
22:05:42 <ais523> because it's as green as it is red, and it's bluer than that
22:05:46 <ehird> then you can move them elsewhere
22:05:50 <ais523> and nor is it very white, grey, or black
22:05:53 <ehird> I find it odd they haven't done that
22:05:55 <ehird> maybe they plan to
22:06:15 <ais523> ehird: there's evidence they plan to, at least
22:06:19 <ehird> well, the folder actually opened would be downloads:/
22:06:21 <AnMaster> here it would be (loosely translated): "master program" (bleh for Swenglish) but the "networking in java" would be called "kurs" which according to dictionaries means "course"
22:06:25 <ais523> there's a setting for download folder in one of the conf files somewhere
22:06:31 <ehird> but it'd show files from ~/.gnome2/downloads
22:06:37 <AnMaster> ais523, which explains why I was confused why it was so long
22:06:50 <ehird> ais523: just the centralised download location or the Nautilus thing too?
22:07:12 <fizzie> How confusing; around here a "course" is a single-topic lectures-and-so-on usually-ends-in-an-exam thing, and a "module" is one of a couple of multi-course building blocks which together make up the degree.
22:07:16 <ais523> ehird: Nautilus seems not to know of it
22:07:22 <ais523> in fact, nothing seems to download there except manually
22:07:29 <ehird> ais523: I invented it
22:07:38 <ais523> although, I think I borked the config when I accidentally renamed every file in my home directory
22:07:39 <ehird> ais523: read up to "incidentally, I find it odd that Gnome has a bunch of disparate download folder settings, and each app does its own download manager" onwards for a few lines
22:07:52 <ehird> ais523: I explain a more Gnomey system that I'm surprised hasn't been adopted already
22:07:52 <ais523> it wasn't nearly as bad as deleting the thing by mistake, but it certainly caused a lot of chaos
22:08:07 <ais523> especially as the desktop couldn't find ~/Desktop, so defaulted to ~
22:08:14 <ais523> and the desktop contained everything in my home folder
22:08:18 -!- kar8nga has quit (Remote closed the connection).
22:08:23 <Deewiant> ais523: Lavender is pretty close to ddddff and it's a shade of purple
22:08:38 <fizzie> There's that ECTS standard "credit" for amount-of-study; can't seem to find out whether they've standardised any related terminology.
22:08:48 <ais523> Deewiant: I suppose colour names are almost impossible to standardise across people
22:08:53 <ais523> to me, ddddff is blue by definition
22:09:02 <ais523> and I base my recognition of colours on that definition
22:09:14 <ehird> blue is 0000ff :-P
22:09:15 <Deewiant> Wikipedia says lavender is e6e6fa, which is more green/red and less blue
22:09:20 <ehird> but yes, ubuntu 6.10 was lavender
22:09:38 <ais523> what colour would you say indigo was, by the way?
22:09:48 <ais523> ever since people said there were 7 colours in a rainbow
22:09:55 <ais523> indigo's allegedly some sort of dark blue
22:09:58 <ais523> but there isn't dark blue in a rainbow
22:10:04 <ais523> all the colours are much the same brightness
22:10:07 <fizzie> "In U.S. education, a course is a unit of teaching that typically lasts one academic term, is led by one or more instructors (teachers or professors), has a fixed roster of students, and gives each student a grade and academic credit.-- Courses are made up of individual sessions, typically on a fixed weekly schedule." That sounds like our courses.
22:10:15 <ehird> depending on the shade
22:10:18 <ehird> either blue or purple
22:10:32 <ais523> fizzie: if US and UK used the terms differently, it would explain the confusion
22:11:06 <ehird> really, it's rather obvious from the years
22:12:26 <ehird> incidentally, one of Epiphany/WebKit's current flaws is that it has no extensions
22:12:36 <ehird> maybe I'll install Ubuntu epiphany-gecko and see if it can do the search bar thing
22:13:09 <ais523> http://blog.rlove.org/2005/10/with-little-help-from-your-compiler.html <-- ouch
22:13:17 <ais523> many people are complaining about the semantic changes depending on compiler in that
22:13:29 <ais523> what I noticed was that they were messing with double-underscore identifiers they weren't allowed to use
22:13:39 <ais523> #define __pure can literally do /anything/ in the C standard, for instance
22:13:41 <fizzie> Let's see what OED says. Wow, that's a long entry; that word is so very overloaded. Full of stuff like "b. Eccl. The prescribed series of prayers for the seven canonical hours." but nothing about education-related terminology.
22:14:02 <ehird> ais523: what's wrong with that post?
22:14:05 <ais523> I can even imagine a legit non-DS9K compiler in which that sort of thing causes link errors
22:14:10 <ais523> ehird: the reddit article explains half the problems
22:14:13 <ehird> argh, I hate xchat-gnome; like xchat, it has absolutely no line spacing between lines
22:14:18 <ehird> reading IRC is quite a chore...
22:14:22 <ais523> but basically, look at __packed, for instance
22:14:27 <ehird> ais523: you didn't link to the relevant reddit thing
22:14:27 <ais523> that makes a semantic change to the code
22:14:35 <ais523> http://www.reddit.com/r/programming/comments/9mg4l/with_a_little_help_from_your_compiler/
22:14:47 <Deewiant> ais523: "For instance." Any others/
22:15:24 <ais523> any of them could bork system headers
22:15:27 <ais523> the #define inline in particular
22:15:33 <fizzie> ehird: "You can always make your own font-variant which includes the line-spacing you want."
22:15:50 <ehird> fizzie: is that a quote :D
22:15:50 <ais523> also the define on __const
22:16:04 <ehird> fizzie: say it again so it is!
22:16:13 <fizzie> ehird: "You can always make your own font-variant which includes the line-spacing you want."
22:16:27 <ehird> fizzie: oh man, what idiot said that?
22:16:42 <ais523> also, the likely and unlikely macros are crazy, in that the gcc versions cast to boolean, the non-gcc versions leave it as an int
22:16:42 <fizzie> ehird: It was this fizz-erly-something guy.
22:17:01 <ehird> fizzie: what a moron!
22:17:18 <ehird> ais523: what's boolean defined as?
22:17:29 <ais523> which casts to boolean whether it's defined or not
22:17:37 <ehird> [[Meh, "likely/unlikely" pale in comparison with the __assume intrinsic of VC++. It's for when you assert and mean it, like a real man! It allows you to bend software reality any way you want! Say __assume(x == 4) and the compiler would eliminate the variable by constant propagation. Say __assume(0) and the compiler would consider the containing piece of code unreachable and eliminate it.]]
22:17:44 <ais523> "#define __malloc" is great, too
22:17:54 <ehird> "whether it's defined or not"?
22:18:00 <ais523> whether boolean is defined or not
22:18:06 <ais523> as in, NetHack has a typedef something boolean
22:18:10 <ais523> and many other codebases do too
22:18:18 <ais523> and splint lets you tell it what your boolean type is called
22:18:39 <fizzie> I'm not sure you can say "cast to boolean" when they actually don't do *that* much to the type, just clamp the values to 0, 1.
22:18:56 <ais523> well, that's the same thing as casting to (C99's) boolean, isn't it?
22:19:18 <ais523> the fun thing about "#define __malloc" is that __malloc is an /incredibly/ likely token to appear in system headers
22:19:18 <ehird> hmm... Pidgin is a terrible IRC client, but I wonder if libpurple(the library underneath)'s IRC support is any good?
22:19:27 <ehird> I hope not, because I like inventing the wheel :P
22:19:32 <fizzie> Isn't !!1L still a long, while (boolean)1L might not be? I don't really know, that's just what I'd think.
22:19:34 <ais523> and deleting it at random could create all sorts of fun
22:19:41 <ehird> although I like reinventing it too, really
22:19:48 <ehird> btw, "libpurple(the library underneath)'s"
22:19:55 <ais523> fizzie: both resulting values will cast to the same values if cast to the same type, though
22:19:57 <ehird> anyone run into a similar syntactical quandary before?
22:20:07 <ais523> ehird: you can do that in email addresses
22:20:44 <fizzie> Yes, but it makes a difference if you put them to the X in "long foo = INT_MAX + X", I guess?
22:20:48 <ais523> __assume would be easy enough to implement in C-INTERCAL, anyway
22:20:56 <ehird> "And a C or C++ program that names a type something_t is also not portable as _t is reserved either by the C standard or by POSIX."
22:20:59 <ais523> although it would have to be an operator, because the optimiser doesn't cross statements yet
22:21:06 <ais523> (predicting flow control is far too difficult in INTERCAL)
22:21:08 <ehird> ais523: that answers my question about whether I should name my types like that in the Splinted program...
22:21:25 <ais523> yep, _t is reserved by POSIX unconditionally for future expansion
22:21:34 <ais523> and by C99 for certain patterns before the underscore
22:21:46 <ehird> hmm... alt-tab has an annoying lag in metacity
22:22:04 <ais523> also, you can't have a function called is followed by a lowercase letter followed by anything, or str followed by a lowercase letter followed by anything
22:22:50 <Deewiant> But you can have a function called is or str followed by a lowercase letter followed by nothing?
22:23:22 <fizzie> The _t typedef naming is pretty common; people like to live dangerously.
22:23:36 <ehird> Nothing is in anything, isn't it?
22:24:27 <Deewiant> In that case, "followed by anything" means nothing.
22:24:49 <ehird> No, it means that folllowed by anything means some or no text.
22:25:21 <ehird> Omitting it implies "end of text"; i.e., strx is illegal but strxfoo isn't.
22:26:48 * ehird wonders if you can gpg-encrypt a loopback device as an alternative to truecrypt :-D
22:27:26 <ehird> like, mount a two-way FIFO
22:27:35 <ehird> and have it gpg encrypt/decrypt
22:27:46 <ais523> <MSDN> # define ASSERT(e) ( ((e) || assert(__FILE__, __LINE__) )
22:27:48 <fizzie> That doesn't sound very seekable.
22:27:51 <ais523> is it me, or is there something very wrong with that line?
22:28:01 <ais523> or does VC++ use a nonstandard definition of assert?
22:28:52 <ehird> ais523: overloaded?
22:28:57 <ehird> this is C++, after all
22:29:11 <ehird> fizzie: well, it decrypts it all at first
22:29:14 <ais523> assert's defined as a macro in the standard headers
22:29:20 <ehird> fizzie: although I guess that's keeping it in memory
22:29:25 <ehird> so what if it's slow! :D
22:29:31 <ais523> that are defined by the standard
22:29:33 <ais523> you can't overload macros
22:29:37 <ehird> ais523: and that's defined in the standard headers, so they're exempt.
22:29:52 <ehird> ais523: clearly it's defined before #define assert()
22:29:52 <ehird> and assert is their function defined before
22:30:03 <ais523> that was a code example
22:30:05 <ehird> #define assert(x) foo
22:30:09 <ais523> that definition of ASSERT isn't a standard one
22:30:12 <ehird> that doesn't use assert, does it?
22:31:17 <ais523> <MSDN, discussing assert> Evaluates an expression and, when the result is false, prints a diagnostic message and aborts the program.
22:31:21 <ais523> well, that's the /correct/ definition
22:31:38 <ais523> so I have no clue what's going on here
22:32:19 <ais523> although they suggest that code should use System::Diagnostics::Debug::Assert instead
22:33:03 <fizzie> It's just a code example, it might be, you know, wrong. Or trying to show how to implement an assert macro (called "ASSERT") with file-and-line-printing, using a custom underlying assert() function, without thinking about the standard assert at all.
22:33:12 <ais523> my guess is that it's wrong
22:33:19 <ais523> it's still shoddy behaviour, though
22:33:36 <ais523> and even if it's meant to be a custom assert, it's incredibly bad style and thus shouldn't be used in code examples in the official documentation
22:33:38 <fizzie> I guess you are allowed to name a function assert() if you want, and call it too, as long as you don't include <assert.h>?
22:37:27 <ehird> ok, firefox crashed
22:37:43 <ehird> time to install epiphany-gecko because I'm too lazy to add the PPA for now and this might have useful extensions
22:38:49 <ehird> ais523: do you know if apt-get purge wipes gconf things too?
22:39:08 <ais523> as in, I know that I don't know
22:39:11 <ais523> not I don't know whether I know
22:39:20 <ais523> but if I'd said "no", that would also be ambiguous
22:40:20 <ehird> ais523: another problem I have with Epiphany is that it uses Gnome default settings for the default font size; in my case 10pt
22:40:28 <ehird> but the default size on the web tends to be 12pt = 16px
22:40:42 <ehird> so <font size=-1> and other pages using sizes relative to the default have small text
22:40:48 <ehird> which is intensely irritating
22:40:54 <ehird> and applies to quite a bit of pages
22:41:01 <ehird> there used to be an option to change this
22:41:07 <ehird> but it seems they removed that, which is a fun regression
22:43:07 <ais523> websites saying smaller-than-default font for everything are insane
22:43:10 <ais523> IMO, it's the website's fault
22:43:27 <ehird> what, for sizing text?
22:43:40 <ais523> for putting a relative size on text that doesn't leave most of the text on the site at default
22:43:43 <ais523> it's default for a reason!
22:44:19 <ehird> Yes, and that reason being that firstly it's a fucking pain to change it and secondly almost all users won't bother changing it even if they can.
22:44:28 <ehird> Oops, no, wait, that's an argument for a browser being PRACTICAL.
22:44:33 <ehird> Besides, I prefer the 16px font on the web
22:44:47 <ehird> But I can't change it without changing the font size for ALL documents, including IRC, because it's the DEFAULT SIZE.
22:44:58 <ais523> you prefer websites and nonwebsites to have differently sized text?
22:45:16 <ehird> For a lot of things, yes.
22:45:26 <ehird> Surprisingly enough, most applications are unlike a web browser.
22:45:27 <ais523> it's an attitude I find hard to understand
22:45:36 <ais523> it's text in both cases, and in both cases generally I want to read it
22:45:40 <ehird> because you barely use any programs
22:45:44 <ais523> so I want it at a good size for me to read
22:46:01 <ehird> no, it's not "text" in both cases
22:46:03 <ehird> it's not an article
22:46:15 <ehird> "Document" in Gnome is almost everything in the window body
22:46:25 <ehird> apart from UI elements
22:47:54 <ais523> ehird: so suppose you were reading an IRC log in your browser
22:48:00 <ais523> would you want that a different size from talking on IRC?
22:48:13 <ehird> Surprisingly enough, it's called a default because it's not appropriate in ALL cases.
22:48:29 <ehird> Oops, wait, in ais523's idealised web nobody sets the font size because the default is always right.
22:48:55 <ais523> ehird: I'm sort of in favour of semantically correct web pages with no stylesheets
22:49:12 <ais523> although I can't get away with it because IE's default stylesheet is (or at least used to be) truly awful
22:49:24 <ehird> So basically, you want to codify every type of information, every context it might be presented in, and just about everything in one XML schema.
22:50:11 <ais523> I don't think the correct appearance of information tends to correlate very much with what that information is
22:50:36 <ais523> as in, reading an email is much the same as reading an essay
22:51:43 <ehird> maybe I should become dictator and then force everyone to read one typography book and one information design book
22:52:03 <ehird> pros: world becomes a better place
22:52:07 <ehird> cons: moral code violated
22:54:34 <ehird> hmm, the Epiphany in the Ubuntu repositories has the required settings to untiny the fonts
22:54:59 <ehird> they're set to the right thing by default, except you have to move it one down and one up
22:55:13 <ehird> presumably because it's filling it in without having saved it yet, and the backend doesn't use the default as the default shown there...
22:58:36 <ehird> ais523: do you know how to delete a gconf folder?
22:58:43 <ais523> I hardly mess with gconf
22:58:54 <ais523> if I wanted to go around endlessly configuring things, I'd use KDE
23:00:13 <ehird> yeah, uh, thanks for the dismissal but i was trying to remove the old epiphany prefs
23:00:26 <ehird> just rm -rfing the dir got rid of the preferences, but left the folder
23:00:38 <ehird> I guess gconf-daemon isn't synchronized yet
23:00:44 <AnMaster> reading up on IA64 asm coding: conclusion is IA64 really rocks
23:00:52 <ehird> AnMaster: it really doesn't
23:00:58 <ehird> considering that it turns out to be slower
23:01:30 <AnMaster> ehird, how much is due to it being so different that usual compilers didn't work very well. Because that is one known issue with it.
23:01:42 <ehird> no, it was just really slow in general.
23:02:00 <AnMaster> ehird, but predicated execution! And lots more.
23:02:40 <AnMaster> ehird, also why? some fundamental deign flaw or just poor implementation?
23:03:01 <AnMaster> (on the hardware level I mean)
23:03:20 <ehird> Considering how much money Intel put into it and how much marketing and the fact that they've made a bunch of Itanium-based processors, and they're all slow, I conclude that the architecture sucks.
23:03:59 <ehird> It's just slow. In general. At everything.
23:04:08 <AnMaster> ehird, C code for benchmarking?
23:04:51 -!- Pthing has joined.
23:04:53 <AnMaster> huh? I'm just asking about the details for how it was deemed slow
23:04:53 <ais523> ARM were claiming to get something like 5 times the performance for the same power usage recently
23:04:58 <ais523> although I don't know what those figures are based on
23:05:05 <ais523> and it's almost certainly been marketing-spun to some extent
23:05:28 <ehird> AnMaster: Because it was slow. at. everything.
23:05:43 <ehird> It's like asking what code was used to conclude that the Pentium 4 is slower than the Core 2.
23:05:51 <ehird> It's not slow at some benchmark. It's just slow in general.
23:06:12 <ehird> ais523: wrong window.
23:06:13 <AnMaster> ehird, How was it messured. I mean, looking at the IA64 architecture it would be best at very parallel tasks, with few dependencies
23:06:28 <ais523> ehird: no, it was a joky example of how to benchmark processors
23:06:28 <ehird> AnMaster: How was WHAT measured?
23:06:30 <ehird> IT IS SLOW IN GENERAL!
23:06:34 <AnMaster> rather than stuff where each calculation depends on the previous part
23:06:37 <ehird> At everything! By multiple people!
23:06:51 <AnMaster> ehird, yes, but what was the specific benchmarks
23:07:03 <AnMaster> as far as I can see it is basically SIMD most of the time
23:07:33 <ehird> In AnMaster's world, if someone says "the processors turned out to all be slow", that means they're referring to a single run of benchmarks by a single group as opposed to it JUST BEING SLOW.
23:08:03 <AnMaster> ehird, I'm wondering what sort of reports you will write at uni when you get there
23:08:52 <ehird> ais523: am I being insane or is he responding to "Itanium turned out to be slower than x86 in every processor they've made" with "You will fail at university because you aren't providing a meticulous citation to a single source of a single benchmark run"
23:09:33 <ais523> AnMaster: I think ehird's trying to say that eir opinion isn't rigorous, just an estimate based on eir own experience
23:09:44 <ais523> if the gaps as large as e claims, an estimate will easily be good enough
23:09:46 <ehird> that's not what I'm saying at all
23:09:53 <ehird> I never said anything close to that
23:11:12 <AnMaster> ais523, anyway, the concept sounds good, possibly the implementation (the actual architecture) was sucky. basic stuff like predicated execution, VLIW, and so on is all great
23:11:24 <AnMaster> plus a lot more money has been spend on x86
23:12:48 <AnMaster> less work often means worse results. Not implying that IA64 had a small budget or effort wasn't spent on it. Just saying overall, x86 had a lot more of that during the years
23:14:50 -!- FireFly has quit ("Later").
23:18:23 <Ilari> How fast CPU is is affected by lots of things. Clock frequency, pipeline depth, auxillary function units, proneness to hazards, etc...
23:18:47 <ehird> Ilari: what benchmarks proved this?
23:18:54 <Ilari> Netburst P4 was bad at pretty much everything except clock frequency.
23:18:55 <ehird> i want specific benchmarks!
23:18:59 <ehird> oh, you don't have any?
23:19:02 <ehird> guess you fail at life
23:21:13 <Ilari> Longer pipeline slows things down by being so long that one can't get enough paralelism to keep the CPU busy. Also, cost of triggering hazard is proportional to length of pipeline.
23:21:28 <ais523> pipelines have a weird effect on efficiency
23:21:37 <ais523> they increase latency, increase throughput, and make mispredictions worse
23:22:53 <Ilari> Yeah, there is some sweet spot where performance enchancement due to higher clock frequency and de-enhancement due to insufficient paralelism and hazards balance...
23:24:16 <Ilari> But at given clock frequency, lengthening the pipeline tends to be negative impact...
23:26:57 -!- ais523 has quit (Remote closed the connection).