Pugs 6.2.8 released! | pugscode.org | pugs.kwiki.org | paste: sial.org/pbot/perl6 | www.geeksunite.net
Set by autrijus on 13 July 2005.
mugwump ?eval all((1...).map:{$_*2 - 1}) 00:11
muhahaha!
evalbot6 pugs: out of memory (requested 1048576 bytes)
acme hugs evalbot6 00:14
geoffb & # time for dinner, methinks 00:24
svnbot6 r5792, fglock++ | * implemented iterator() in Span/Recurrence/Set-Infinite, added a few tests 02:35
coral ?eval <1...>.pick 03:04
evalbot6 Error: pick not defined: VStr "1..."
coral ?eval <1..100>.pick
evalbot6 Error: pick not defined: VStr "1..100"
geoffb ?eval 1..100.pick 03:22
evalbot6 Error: pick not defined: VInt 100
geoffb ?eval (1..100).pick
evalbot6 20
geoffb ?eval (1..100).pick
evalbot6 17
geoffb Man, I am just loving forth . . . .
at least, as much as I've gotten to in the tutorials so far 03:23
Reminds me of my old assembly language days
svnbot6 r5793, Stevan++ | Perl6::MetaModel - (p5 version) fixing up a few little things 03:35
geoffb stevan, don't remember if I asked you this already, but: 03:38
Have you gotten to the point in the MMs that remaining significant changes will be due to @larry doing stuff to the language? 03:39
ingy seen autrijus 03:46
jabbot ingy: autrijus was seen 2 days 8 hours 4 minutes 34 seconds ago
stevan geoffb: no, there is still some stuff to do, namely roles attributes 03:52
geoffb: currently roles only support methods
geoffb k 03:53
thx for the update, stevan 03:54
stevan geoffb: I am sure there are other missing peices too, I have to re-read AES12 again 03:55
geoffb fair enough 03:56
clearly forth is making sense to me, since I am finding bugs in the sample code in the tutorials . . . :-)
stevan but the basic structure is there :)
geoffb stevan++ # many days of hard work 03:57
stevan geoffb: thanks :) 03:58
svnbot6 r5794, Stevan++ | Perl6::Code -
r5794, Stevan++ | * named subs work (recursive too)
r5794, Stevan++ | * named multi-subs work (also recursive)
r5794, Stevan++ | - NOTE: these are using very primative MMD based on the number of args
geoffb wheee, jforth tutorial chapters down, starting on pforth tutorials 05:08
Well, that was quick -- the pforth tutorial appears to be a subset of the jforth tutorial . . . 05:11
coral heh 05:13
ingy hi stevan 05:17
Khisanth mm assembly
geoffb Khisanth, oh yeah 05:20
wee, fun sidetrack into color forth; on to Forth Meta Compilation 05:32
.oO( I will traverse the forth branch and pop back to harrorth soon, I swear ... er, to myself, I guess )
 05:33
Daniel_Nee Dear Autrijus: I'm waiting for you at S-Team's office, thanks. 06:19
geoffb seen autrijus 06:20
jabbot geoffb: autrijus was seen 2 days 10 hours 39 minutes 8 seconds ago
svnbot6 r5795, masak++ | Minor spelling fix in ext/Set-Infinite/ChangeLog 06:32
Daniel_Nee Dear Autrijus: We are supposed to work on the nanshan RDS project in Monday afternoon and I'm waiting for you at S-Team's office now, thanks a lot. 06:56
szabgab hi nothingmuch 07:03
geoffb haskell tutorials down, forth tutorials down, chapter 2 of harrorth down, and now . . . 07:05
geoffb is going down for a nice long sleep.
& # sweet, sweet slumber 07:06
Darren_Duncan seen autrijus 07:39
jabbot Darren_Duncan: autrijus was seen 2 days 11 hours 58 minutes 2 seconds ago
Darren_Duncan I noticed he hadn't updated his journal in 5 days ... perhaps nothing interesting going on? 07:40
anyway, good night
dada ?eval [>>+<<] (1,2,3); 10:30
evalbot6 Error: Hyper OP only works on lists
dada ?eval >[+]< (1,2,3);
evalbot6 Error: unexpected ">" expecting program
dada ?eval >>[+]<< (1,2,3); 10:31
evalbot6 Error: unexpected ">" expecting program
dada doesn't recall the syntax for the reduce meta-op
?eval [>+<] (1,2,3); 10:34
evalbot6 Error: unexpected "[" expecting program
dada ?eval [+] (1,2,3);
evalbot6 6
dada oh, that was simple
?eval [+] (1,2,3,4,5); 10:35
evalbot6 15
dada ?eval [,] (1,2,3); 10:38
evalbot6 (1, 2, 3)
dada ?eval [,] (1,2,3,4,5);
evalbot6 (1, 2, 3, 4, 5)
dada ?eval reduce( { $^1 + $^2 }, (1,2,3)); 10:39
evalbot6 6
dada ?eval reduce( { $^1 , $^2 }, (1,2,3));
evalbot6 ([\1, \2], \3)
dada ?eval reduce( { ($^1 , $^2) }, (1,2,3)); 10:40
evalbot6 ([\1, \2], \3)
dada ?eval reduce { [,] ($^1+1, $^2+1) }, (1,2,3); 10:46
evalbot6 (3.0, 4)
dada ?eval reduce { ($^1+1, $^2+1) }, (1,2,3); 10:48
evalbot6 (3.0, 4)
dada ?eval reduce { ($^1+1, $^2) }, (1,2,3); 10:49
evalbot6 (3.0, \3)
dada ?eval reduce { ($^1, $^2+1) }, (1,2,3);
evalbot6 ([\1, 3], 4)
dada ?eval map { $_+1 } (1,2,3);
evalbot6 (2, 3, 4)
dada ok, question for the functional warriors out there: is it possible to define "map" using "reduce"? 10:53
12:35 sbkhh is now known as Odin-
svnbot6 r5796, iblech++ | * Usual svn properties. 12:54
r5796, iblech++ | * PIL2JS:
r5796, iblech++ | * Fixed %hash.delete(...).
r5796, iblech++ | * Fixed handling of arrays where some elems were .delete()d.
r5796, iblech++ | * Added support for numification and stringification of hashes.
r5796, iblech++ | * New builtins: &key, &value, &keys, &values, &kv, &pairs (but &pairs doesn't
r5796, iblech++ | work yet correctly, will investigate).
r5796, iblech++ | * Unbroke &postcircumfix:<{ }> for hashes.
r5796, iblech++ | * @a["23"] works now (i.e., it means @a[23]).
r5797, philcrow++ | Added Newton.pm to algorithms examples to show off Currying and 13:00
r5797, philcrow++ | parameter defaults.
13:26 joepurl_ is now known as joepurl 13:47 viirya_ is now known as viirya
Limbic_Region perlmonks.org/index.pl?node_id=477815 # Perl6 MetaModel question 14:05
seen stvn 14:06
jabbot Limbic_Region: I havn't seen stvn, Limbic_Region
Limbic_Region seen stevan
jabbot Limbic_Region: stevan was seen 10 hours 7 minutes 56 seconds ago
stevan hola Limbic_Region 14:10
svnbot6 r5798, Stevan++ | Perl6::MetaModel (p5 verison) 14:11
r5798, Stevan++ | * added 'build' traits to attributes (has $answer = 42)
r5798, Stevan++ | - added test file for this
r5798, Stevan++ | * added the CALLONE and CALLALL methods (desugared $obj.?meth(@args), $obj.+meth(@args) and $obj.*meth(@args))
r5798, Stevan++ | - added tests for these
Limbic_Region speak of the devil
stevan ping
stevan Limbic_Region: thanks for the heads up :) 14:12
masak what's the best way to help if you're eager for bugs (for which there are already failing tests) to be resolved? :) 16:38
Aankhen`` Learn Haskell and fix them. 16:54
Khisanth heh
Aankhen`` # which I have not yet tackled myself :-\
Or perhaps bribe autrijus. =)
ingy stevan: are you going to oscon?
Limbic_Region August 1st is right around the corner :-( 16:55
Khisanth Aankhen``: s/bribe/sponsor/ :)
kinda the same but not quite :)
ingy for that matter, does anyone know which lamdacamels are attending oscon? 16:56
Khisanth but learning Haskell would be good anyway
masak is darn close to learning haskell just to fix his bugs
Khisanth DO IT! :) 16:57
s/^/JUST /
masak hides in a corner, studying haskell
ingy just realizes that unordered lists are really ordered after all... 16:58
Khisanth significant whitespace doesn't seem so bad with Haskell mode in emacs
masak is there any way to run perl cgi without a server? 16:59
QtPlatypus You can sort of do it from the command line.
masak QtPlatypus: sounds good. go on. 17:00
Khisanth also plenty of HTTPD modules :)
masak sounds not-so-good :)
Khisanth what are you trying to do? 17:01
masak i'm trying to set up a perlnomic sandbox 17:02
Khisanth you could run any CGI program from the command line, it's just that the output will be going to stdout
masak hm, maybe that's all i need
i can always redirect to file 17:03
stevan ingy: no I am not going to OSCON, YAPC was my only conference this year
Khisanth of course if you are using cookies and other things, it's probably not the way to go
masak i don't think i am 17:04
however, i am using url parameters sometimes 17:05
hm... running from command line works 17:09
is there an easy way to send parameters into the cgi script from the command line? 17:11
(i realise that there may be more appropriate fora for such a question)
Khisanth that depends ... is it use CGI.pm?
masak yes
Khisanth you could supply it on the command line as well 17:12
masak really? how?
Khisanth ./foo foo=bar
it used to tell you how when you ran it on the command line :/
masak Khisanth: it works! thanks! Khisanth++ 17:13
CGI.pm++
geoffb ingy, I will be going to OSCON. 17:15
masak, funny, I'm learning haskell precisely because I have the same issue (bugs that are, well, bugging me)
Khisanth masak: what bug are you trying to fix? 17:17
stevan geoffb: interested in doing a Haskell port of the MetaModel? 17:46
it would be an excellent learning tool :) 17:47
masak Khisanth, two right now 17:49
geoffb stevan, good point -- let me get through harrorth, then I can consider that. 17:50
:-)
stevan geoffb: ok
masak the first is a junction bug (t/junction/array_deref.t) which prevents me from tidying up examples/games/tic_tac_toe.p6
the second is a constructor initialized member shifting bug (t/pugsbugs/attribute_list.t) that puts wizard.p6 in its infinite loop 17:52
clearly, i'm into pugs because of the games :)
geoffb :-)
geoffb is as well -- because he wants sdl_perl working well under pugs 17:53
masak geoffb: we both have the same itch (bugs), path (harrorth) and goal (fixing pugs)... i feel a bit less alone in the world :) 17:55
geoffb :-)
ditto
masak geoffb: haskell is some kind of logician's voodoo, isn't it? 17:56
i think i grok the easy parts 17:57
geoffb masak, I think so . . . I will say this -- it really seems to be grinding too hard on that axe
masak, ditto
masak pattern matching, currying, all that stuff
but i still tense up as soon as someone mentions monads
geoffb And . . . this just clarified in my mind now . . . I feel way too much of that C++ "you think you're just adding two things, but really you've just initiated several billion steps behind the scenes" issue 17:58
masak i don't know why, but passing the world around as a parameter scares the heck out of me
geoffb heh 17:59
masak maybe i should just try using monads in a meaningful, practical way 18:00
and see who masters whom :)
geoffb The part that scares me is knowing so much is being carried around, but never changed (only operated on) -- the GC must be hell
heh
masak yes, no wonder gc and functional programming go hand in hand 18:01
fp is both more logical than imperative programming, and further removed from common sense -- it's strange 18:02
geoffb :-) 18:03
masak i actually tried to fix a pugsbug a while back
it looked like an easy one
Odin- Strict logic is usually very far from common sense. 18:04
:>
masak but changing the source code only yielded typing errors from ghc
good thing there's svn revert
:)
geoffb just translated a perl5 one-liner into forth
masak Odin-: seems so
geoffb I feel all code-studly now
masak geoffb++ # for speaking forth 18:05
Khisanth geoffb: how many line of forth?
geoffb forth++ # for just being fun
Odin- Which is why Perl sometimes irritates functional folks ... Perl has traditionally put the focus on common sense, ignoring the "real" logic... :p
geoffb Khisanth, the core is two words and three variables
Khisanth forth seems stranger than Haskell by several magnitudes :) 18:06
masak Odin-: which is why perl6 is such an interesting combination
geoffb I had several debugging words to help while I figured out why it was broken at first
Khisanth, depends on where you came from. For me, it makes MUCH more sense than haskell 18:07
masak for me too
Odin- masak: Yeah, although I'd argue it remains closer to its tradition than functionality...
geoffb nothingmuch++ # harrorth forcing everyone to expand their horizons/mind
Odin- (Now, THAT was a nicely ambiguous sentence. :) 18:08
Khisanth haskell makes sense, just need to get the syntax down :)
masak Odin-: that's the beauty. perl6, akin to the borg, looks at fp and sucks up the good parts
Khisanth: then i look forward to the day when i get haskell syntax down :) 18:09
geoffb Without telling me that an array can't contain both a char and an int, dammit.
Khisanth geoffb: what language? :)
geoffb Khisanth, I meant -- P6 doesn't enforce the limitations (that I don't like) that haskell does 18:10
masak geoffb: haskell needs those limitations, for strictness
it's part of its game
Khisanth well most languages do :)
geoffb is somewhat biased about type theory -- all I want it for is to get the compiler to do optimization for me. The rest, I don't want.
Khisanth an array of tuples! :) 18:11
masak not many languages do as much typing as haskell
geoffb Khisanth, I tried that -- hugs complained that the tuples were not all the same type. :-)
masak :)
same problem, one level down. neat 18:12
Khisanth geoffb: perhaps we are not thinking of the same thing :)
I was thinking of what would be an array of structs/unions in C :)
geoffb Khisanth, that wouldn't be too surprising, given that I've been reading haskell for all of about half a week 18:13
Khisanth geoffb: I have been reading much less :)
perhpas there is some limitation I didn't get to yet
geoffb: however, one of the first few YAHT exercises dealt with that 18:14
geoffb Well, I thought about using tuples with a lot of entries, to act as a union, but tuples are fixed length (and all tuples within the list have to have the same length) -- which means that if you want to be able to store any type in the list, you start writing functions that take a list of tuples of length n and producing a list of tuples of length n + 1 -- and pretty soon you start hitting the wall that many library functions are d 18:15
efined only for short tuples
Ah, YAHT is the one HT on the list that I haven't read through -- I needed a break after the first three or four haskell tutorials. 18:16
& # life calls 18:17
masak geoffb: say hello to life from me 18:19
geoffb masak, heh 19:13
cognominal seen autrijus? 19:17
jabbot cognominal: autrijus was seen 2 days 23 hours 36 minutes 28 seconds ago
cognominal I became addicted to his journal... 19:18
brentdax Do Perl 6 regexen support escape sequences yet? Because I tried to s{\t}{foo} and it applied the rule to 't's. 19:56
PerlJam brentdax: yes, but AFAIK, PGE only supports \d, \D, \w, \W, \s, \S, \n and \N 19:58
you could easily hack it to support \t and \T though
brentdax What file would I need to edit for that? Wouldn't mind getting away from my own software for an hour or two. 19:59
PerlJam brentdax: probably compilers/pge/PGE/P6Rule.pir
one of the files in that dir anyway
brentdax In Parrot?
PerlJam aye
brentdax Alright. Better update my Parrot first, then... 20:00
Hmm...all the existing backslash sequences seem to map to character classes, rather than individual characters, and to have specialize opcodes. 20:21
Khisanth but \t you mean tab? 20:22
brentdax Yeah.
Khisanth wouldn't this be something to do in pugs not PGE?
brentdax ...maybe? I dunno.
PerlJam Khisanth: he was doing a s///. In pugs that's handled by either pcre (if you've used :P5) or PGE 20:23
unless I'm totally mistaken (it's been a while since I've mucked with it)
Khisanth hmm is it really wrong to expect things to work like perl5? :) 20:24
brentdax Meh, I'll just put a tab in my editor. 20:25
Khisanth hooray for invisible code :)
brentdax Indeed.
brentdax emits a cough that sounds suspiciously like the word "python". 20:26
Khisanth PerlJam: so the Perl 6 compiler isn't suppose to know anything about rules?
PerlJam Khisanth: PGE's jog is to know all about rules. When we've bootstrapped ourselves out of pugs+PGE land into a real perl6 compiler, it will know about rules. 20:28
er ,s/jog/job/
or if autrijus gets it in his head that he can do better than PGE and does. 20:29
Khisanth well ... the reason I ask is because the left side of s/// has had pretty the same features as Perl's double quotes 20:30
so that would mean PGE has to know about Perl's strings or perl6 has to know about the rules does it ?
PerlJam Currently the former by way of parrot ("perl's strings" are really "parrot's strings") 20:31
again, if I'm not mistaken 20:32
anyway, time for me to go preemptively rescue the wife from invading dinner guests
Khisanth oh wait ... I keep forgetting Parrot is suppose to be flexible to the point of being absurd :)
putter ?eval {x => [3,4]}.perl 21:11
evalbot6 '{(\'4\' => undef), (\'x\' => 3)}'
putter sigh. 21:12
PerlJam: one of my background tasks is to write a rules engine on top of pcre. the task is not inherently hard (given that not all of rules can be supported), but it's a maze of time eating stuff like the above. might be working this evening, might not be working for months. sigh. 21:15
masak putter: that's a weird evalbot6 result 21:17
Khisanth looks like a bug :) 21:24
putter: you mean a rules to pcre translator? 21:25
putter :)
yes
Khisanth Perl6::Rules? 21:26
although that isn't pcre :) 21:27
putter I wonder if we need a "state of pugsbugs" periodic summary. The bug was forseeable from known pugsbugs, but it's been a while since I checked, and I had drifted into assuming the whole nested arrays and hashes mess had been straightened out.
yeah, I also have a mutant P6:R which is slowly being updated to reflect current syntax and semantics. not very far along. but perhaps it should be checked in at some point. 21:29
basically when one path has gotten too frustrating to pursue, i push on another. 21:30
Khisanth hmm a p6 compiler written in p5? :)
putter yes... have bits of that too... ;-) upsides included clean api (working stringification and Filter::Simple), downsides include pain getting some regexps to work (and "work" is perl-version dependent), the non-reentrant re engine (so you end up almost as cripled as with pcre - the reason P6:R isnt being maintained(?)), and ... something else I don't recall at the moment. 21:38
putter ponders finding someplace off MANIFEST to drop a load of non-working cruft into vcs... eh. 21:40
oh, big upside is you don't lose hours because something... odd... is happening. I'm quite tempted to do a ruby rules engine for the same reason. 21:42
masak thinks a "state of pugsbugs" summary would be a good thing 21:49
svnbot6 r5799, putter++ | Extended t/pugsbugs/parsing_hash.t to indicate arrays nested in hash refs also misparse.
putter ahhh... t/pugsbugs/attribute_hash.t... that's why (something) wasn't working earler...
masak :)
g'nite, all &
putter & 21:50
?eval [{1=>2,3=>4}].elems 21:53
evalbot6 2
putter I really thought that had gotten fixed. puzzling. 21:54
ah, the pugsbug/ test went away(?), but it's :todo<bug> in t/data_types/mixed_multi_dimensional.t. 21:58
putter thinks again that :todo<bug> should be smoked as dark red rather than dark green.
coral is there a yellow? 22:01
gaal bonus (unexpected success)
hi
putter I wonder if one could integrate t/ and the language spec/documentation, ie "literate programming", so the docs change color when something isnt working...
the color scheme seems targeted at "developers" (ie, a bug known is greenish, red and yellow indicate something unexpected is happening), rather than "users" (ie, three little dark green boxes don't exactly scream "don't uses nested hashes and arrays in your data structures!! or you'll be sorrrrryyyyy"). ;) 22:04
perhaps we need two color schemes?
hi, gaal
gaal putter, i like the colored spec idea :) 22:05
putter :)
gaal there already is some integration
you know, they link to the specs
putter right.
gaal so the thing that mangles the specs can have some smarts implented in it to look for the text around where the actual link is 22:06
but it'll often guess wrong
putter hmm, so rather than literate programming, one could postprocess the docs by looking groveling over the tests' links.
gaal which is done already
look at nothingmuch's smokes, they include docs
putter (really!) goes to look... 22:07
Khisanth hmm so with the appropriate CSS, different sections of the doc could have different colors dependings on status?
Khisanth would hate to read a document in green though 22:08
putter gaal: could you suggest a test which has such docs?
gaal problems: the link specification mechanism must be lightweight otherwise people won't use it
putter Khisanth: :) 22:09
gaal undef.t has a few
vgrep for L<
in nm's smokes they show up as actual links
putter ah, yes, neat. 22:10
gaal as for the specific color scheme, yuval has voiced frustration over everybody having different opinions about this. (as, i guess, co-maintainer) i think welcome patches for alternate css. but i don't know how the actual css alternattion works, so please don't just set *a* different stylesheet. 22:11
s/think/think we would/
and s/set/send/ 22:13
Khisanth hmm but one html document can have multiple CSS :)
gaal yes, if you rig it. please rig it. :)
Khisanth it's just a matter of using a browser that lets you set a default!
gaal wow it is so humid here 22:14
gaal looks for the airco remote
putter actually, I wonder if we really just need some slightly more organized bug bashing. months ago there was the concept of "trade a test for a patch". but that was mostly autrijus, and as his focus shifted to the backend, and we all didn't pick up the slack, that property was lost...
re colors, I've seen sites with a pulldown list selecting the site's css and thus appearance... I'm not sure how it's actually implemented... 22:16
gaal putter: that's indeed a problem, we're reaching the size most people don't know who can fix what or even if something's worth fixing instead of the bug being "refactored away" by a new component 22:18
wolverian <link href="default.css" rel="stylesheet" type="text/css" title="Default" /> <link href="another.css" rel="alternate stylesheet" title="Another" type="text/css" />
not all browsers have a UI to change to an alternate stylesheet - some JS can help there. 22:19
gaal so it's still possible to 'claim' bugs but if i want to petition bug X be fixed i may not know who to moose about it.
maybe we should just remind people to look for bugs to claim. :)
Khisanth how about just switch to "fix everything you can and have time for" :) 22:20
gaal that's kinda what i was thinking. :) 22:21
gaal starts a smoke 22:22
putter wolverian: thanks! googling turned up www.alistapart.com/articles/alternate/
gaal hmm, there weren't any commits in src/ in a while.
Khisanth but there is something to be said for a task seeming too daunting and nobody wanting to start :) 22:23
wolverian putter, ALA is nice.
gaal okay, it's 1:23, looks like time to shower and go to sleep.
putter re bugs, 22:24
there is also dynamics like... "a few sentences from autrijus on how to approach something can save hours of work". 22:25
gaal which are true :)
putter we haven't really (I think?) tried to find a social organization/dynamic that has nice properties... 22:26
gaal k, this is me. good night all :) 22:27
putter g'night &
gaal why, anarchy :)
putter yes, but part of anachy is folks doing entropy reduction tasks. this area seems to be ready for one...
:)
gaal indeed. 22:28
&
putter &
wolverian: re "ALA is nice", I hadn't seen it before. adding it to my bloglist. thanks :) 22:29
brentdax ALA is awesome.
nothingmuch hola 22:40
shyte, how did that happen
nothingmuch can't paste 22:41
nothingmuch.woobling.org/paste.txt 22:42
i tried to reply to allison to see if i know what I'm talking about
can anyone verify?
putter re paste.txt, the couple I know without checking look right... 22:48
putter wonders how much the planned PIL-related changes will change PIL... 22:49
nothingmuch wonders what data structure mutt keeps messages in 22:50
and if it really resorts the whole list when an update to the mailbox happens
putter ouch.
wolverian I've had to switch to muttng because mutt doesn't cache the headers at all
and it's just painfully slow to open a folder with 5000+ mails in it like that
but that might not actually have anything to do with your issue, now that I reread what you said. sorry. :) 22:51
nothingmuch wolverian: i do too 22:52
i keep it running inside screen
30,000 messages
when I hit 700 i said to myself "i really have to sort all this"
wolverian hehe.
I sort my mailing lists.
(I mean, procmail does.)
nothingmuch then it hit 2000 or so and I was like "oh man, now I'll need a whole 2 hours to sort through all this", so i pushed the deadline a bit more
my problem is that if i let procmail sort them I don't get around to reading them 22:53
i mark threads as read usually
if it's a long deathless thread
or something about a technical issue I don't care about
i tried deleting threads for a while
but then I lose context
I'd like a feature where if i delete a thread, children of that thread are automatically deleted for me, until the thread has been dead for a week 22:54
and then it's completely purged
brentdax I'd really like a mail client with Gmail-ish labels. Those things are really far too handy to live without now that I've used them. 22:58
I keep hoping somebody will write a Thunderbird extension for it. 23:00
putter nothingmuch: fyi, 40 min ago there was a discussion of adding alternate style sheet capability to smoke. wolverian provided the basic code, and this www.alistapart.com/articles/alternate/ reference was found. 23:03
nothingmuch wishes he had the time to finish what he started with consolidation
putter: were any decisive conclusions reached? 23:04
putter gaal solicited a patch to add multi-css capability.
nothingmuch was a patch made?
nothingmuch reads backlogs instead of harassing putter 23:05
putter Not yet. I started exploring, but it looks like I've never actually run smoke...
;)
It looks like simply breaking out the css into a file and adding "one" line in it's place. It wont change things for IE users (requires javascript hackery aswell), but that can be a rev 2. 23:07
nothingmuch oh shit 23:08
putter ?
nothingmuch i hate feature creep
they are called cascading for a reason
=(
this is probably much better handled by overridding on a case by case basis 23:09
putter I don't suggest cascading, but having the css in a separate file, and the files in svn, will allow some exploration of color choice. 23:10
wolverian I think the idea is to just allow users to choose between multiple styles, which this is ideally suited for.
nothingmuch well, overriding the css is as short as:
use base qw/Test::TAP::HTMLMatrix/;
sub css_file { "your path to file.css" };
blam
wolverian I define user as the person viewing the report, not the person that creates it 23:11
nothingmuch inline CSS really means "slurp the file and put it in <style> tags"
wolverian so I'd like a method to change the style inside the browser.
nothingmuch wolverian: what if we use existing browser features?
FF can swap css styles when it's provided with several alternatives...
wolverian that's exactly what I suggested. 23:12
nothingmuch i really don't want to introduce more and more features that have nothing to do with actual reporting
oh...
the link discussed javascript and stuff, so i was begining to be scared
wolverian I also pointed out some browsers can't do it without us providing a JS interface
putter This was all prompted my realizing I was screwed, having used array refs nested in a hash ref as a datastructure, when the three little dark green boxes in smoke were wispering ("it just doesnt work...").
nothingmuch okay, so what if we do something simpler:
we let you generate either an HTML snippet, or a full page 23:13
the full page is what we have now
and the snippet is just the unstyled table
and then we can easily create alternatives that don't have the kitchen sink reimplemented in perl 6 compiled to javascript, stored inline in the template ;-)
wolverian or just do the alternative stylesheet <link> method, and don't care about IE users 23:14
nothingmuch because 95% of the feedback i get for TTH is CSS suggestions, not contributions for code or documentation ;-)
putter as I developer, i like the current colors' emphasis on expected vs unexpected. as a user (p6 coder), I want different colors, which scream at me in red when something is buggy, even if it's been known to be buggy for a really long time...
wolverian frankly, I think that's the most viable alternative.
putter when I view your smoke, the colors i need depend on why i'm there...
nothingmuch putter: as a pugs user that causes lots of confusion because lots of the pugs test suite fails =)
but I agree
that's why i want to write a smoke server 23:15
where people submit YAMLs
and users get to generate reports on the fly
putter re "95% of the feedback", having css in a separate file, which seems plausible idea by itself, would allow you to say "create an alterate .css file in mumble/. That will solve your immediate need, and if the changes seem to be popular/I like them, I'll add them to the default.css". 23:18
nothingmuch putter: the CSS *is* a separate file
this is just not the default way the smoke report for 'make smoke' is generated 23:19
so that the user won't have to copy two files around
putter ah.
nothingmuch wow, my bugs submitted to Fire have been reviewed
submitted 2.5 years ago 23:20
i switched to adium because it had proper hebrew support
wolverian hmm. can you use a rel="alternate stylesheet" attribute on a <style> element? that way we could include everything in one file too
nothingmuch and now they to fix hebrew in fire =_
wolverian (admittedly it wastes bandwith..)
nothingmuch wolverian: not yet
wolverian nothingmuch, oh, okay.
nothingmuch patches welcome =
)
wolverian no, I mean, does it work in browsers?
nothingmuch it's 2:20 am and I'm still trying to push out a build
oh, I have no clue
wolverian apparently not. 23:21
ah well.
putter the ALA article (I think?) said something about naming style stuff to allow it to be turned on/off as a group. 23:22
nothingmuch 2:23am up 47 days, 14:41, 27 users, load average: 8.12, 6.74, 5.77 23:23
damnit 23:24
no wonder the compilation is not finished
nothingmuch has seen load averages of 20 this week =(
putter where?
nothingmuch work 23:25
putter ah
nothingmuch we have 2 strong boxes of every kind
i need one of each for the weekly build
putter nifty
nothingmuch and we also have like 15 "strong enough" linux/windows boxes
but people still flock to the strong boxes 23:26
so while linux, aix, solaris and windows have nearly finished running the sanity test suite, hpux is not yet done compiling
putter looks forward to linux having easy to use resource groups.
nothingmuch i think i will liberally apply BOFH skills
geoffb Nice everyone other than you who logs onto the strong boxes . . . assuming you have root access (or a friendly admin) 23:27
nothingmuch not that it'll help much, it's prolly out of ram
geoffb and there you go.
nothingmuch geoffb: yeah, I should have thought of that 3 hours ago
that's what I normally do
sometimes i even sigstop
geoffb heh
nothingmuch if the disk is really being beaten to death (we make a database engine)
geoffb which one? 23:28
(s/one/db engine/)
nothingmuch www.hyperroll.com
special purpose stuff
say you are a company that makes hygene products
you want to know how many hicks bought orange toilet paper in tokyo, split up into monthly and daily sums, over the last 10 years 23:29
this product is supposed to know how to do this stuff fast
putter later & 23:30
nothingmuch ciao
geoffb right, new "entry-level" packages at $35K license for once CPU. That's special purpose pricing, all right. :-)
nothingmuch geoffb: =) 23:32
you know you're really evil if you have to use xargs to make your renicing easy
wolverian xargs++ 23:33
nothingmuch wolverian, putter: some TTH patches i'd like to see: 23:34
alternative styles for CSS supported in the non javascript way 23:35
making the todo reason another class for each cell
so that we can tell apart todo<bug> from something else
but only if the reason is !~ /\s/
or actually =~ /^\w+$/ for that matter
i can't get around to it in the next few days 23:36
but we'll be needing it for work eventually
or rather, i would like to make the work people use it
only a handful do
geoffb finishes reading marketing blurbs on hyperroll website . . . interesting, looks like a SQL*net proxy with some internal smarts 23:39
.oO( I wonder if it just replaces moronic queries, as video drivers commonly recognize and replace moronic shaders )
nothingmuch geoffb: what's sql*net?
geoffb Oracle's wire protocol, IIRC 23:40
nothingmuch oh
the way the product is usually used:
some company paid $$$ to oracle or ibm to store their data in oracle or db2
or occasionally mssql
then company wanted to do funny queries
so they paid $$$ to hyperion to use essbase
but then it was too slow
they bought our product
which basically takes data from oracle/db2 etc 23:41
and OLAPish meta data from essbase
and then does some precalculation
and then you can query
and i think it can also plug the results back into the db by hacking views and stuff 23:42
geoffb Who was it that said that almost all computer science problems come down to exercises in caching?
nodnod, makes sense
At least you guys are in the comfortable position of being able to tell your customers to do it the way you tell them, or they'll be sorry, knowing they just proved it to themselves with the last few stacks of cash they flushed . . . . 23:43
nothingmuch autrijus has a nice saying 23:44
"computers aren't important... the computer is just a level 2 cache between me and the internet"
geoffb :-)
nothingmuch i think i see what you mean 23:45
e.g. our staff configures stuff based on what client says they need
and that's it?
geoffb Having worked at SAP, and then a company that used SAP, I can tell you that customers commonly say "we don't want to do it the way the tool likes", and the tool vendor says "If you try to be contrarian, your user experience will suck." And then the customer does it back asswards, and their user experience sucks . . . so they complain about it. 23:46
hyperroll knows that their customers already learned that lesson. 23:47
Fixing bad user experience being pretty much their reason for existing. :-)
nothingmuch heh
i guess so
although it's not just bad user experiences 23:48
it's a niche product
the integration features exist for the bad user experience
which is most of our client base
but the real smarts of the product is in precalculation
geoffb nodnod 23:49
bleah, I am just not in the mood to study this stuff right now, but I need to.
bah, bah, BAH
nothingmuch need to? why? 23:50
geoffb Because I can't fix pugs until I grok this crap
sorry, getting cranky about monads
And with autrijus hors de combat (sp?), and bugfixing having hit a local minimum, it's up to us that have complaints 23:51
obra seen autrijus 23:52
jabbot obra: autrijus was seen 3 days 4 hours 10 minutes 51 seconds ago
obra wow
geoffb yeah, painful that
nothingmuch oh 23:53
geoffb: tried harrorth
?
geoffb Yes, actually, I was just working through chapter 3 when I got sidetracked into Meet the Mondads
er Monads
nothingmuch it sidetracks on it's own 23:54
geoffb "Somebody's got a case of the Mondads!"
nothingmuch Reader and IO are dissected to death
maybe that can help
i didn't know much about monads when I started writing the Reader chapter 23:55
but by the end I grokked them
geoffb Oh, trust me, I'll read it all. I just don't particularly want to at the moment, but I'm trying to keep my eyes on the prize
nodnod
nothingmuch oh
i thought you were wanted to go on =)
geoffb (and thank you, btw, for creating harrorth -- it's helpful so far)
nothingmuch geoffb: next chapter is going to be lots of forth fun 23:56
but I barely have time for it
geoffb (and it's more fun to read than the tutorials it points to :-)
nodnod
mmmm, Forth.
get with it, you! You need to stay in front of masak and I
nothingmuch with 2 stages of bootstrap i've gotten to a point where most of the system is written in forth 23:57
what we need is the stack and the heap (and inherently pop, push, @ and !)
geoffb schweet
nothingmuch and the bootstrap
and things like math, etc
geoffb :-)
nothingmuch and even the compilation loop can be written in forth on top of it
the boostrap itself is basically: take the prelude 23:58
completely unordered
make a dependency tree
and build the words using a non forth compiler
just enough for ':' and ';' to work
then just evaluate the rest of the prelude in an order that doesn't break
geoffb as long as it is orderable . . . 23:59
nothingmuch well, it is
the leafs of the dependency tree are the primitives
geoffb well, that's good
:-)
nothingmuch there has to be a special case somewhere
or the universe will explode

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!