pugs.blogs.com/dist/Perl6-Pugs-6.2.13.tar.gz released! | run.pugscode.org | spec.pugscode.org | paste: sial.org/pbot/perl6 | pugs.blogs.com
Set by audreyt on 17 October 2006.
audreyt TimToady: so the sequence is actually 00:00
BUILDALL -> BUILD/CREATE -> BUILD/CREATE -> BUILD/CREATE
to agree with what we've said above
otoh, maybe CREATE can have knowledge of all attributes of derived classes, and "mask" the irrelevant parts 00:01
so it's CREATE -> BUILDALL -> BUILD (part of the layout invisible) -> BUILD -> ... 00:02
that however means all intermediate classes can share one single layout
which is probably not the case for cross language classes.
TimToady the layouts have to negotiate that...
generally by not negotiating it...
audreyt hmm? 00:03
TimToady to the first approximation, shared repr is only possible within a given layout
different reprs are assumed to delegate instead
00:03 lambdabot joined
TimToady meeting, so I'll be a bit distracted... 00:04
audreyt but then itneeds to fire CREATE multiple times
i.e. the first BUILD/CREATE strategy above
otherwise deleegation can't be setup
TimToady presuming that outermost CREATE knows everything that needs to be done. 00:05
audreyt actually each stage of CREATE is a simple layout-compat test
if compt, merge
if not, setup delegation 00:06
either way, go to next BUILD
and so on
otoh, if outermost CREATE knows everything, then it can just create one single candidate that has the neccessary heterogenous layout 00:08
CREATE (CREATE+CREATE+CREATE) -> BUILDALL -> BUILD (part of layout invisible) -> BUILD -> BUILD ...
I guess I like it more.
audreyt codes it up
another thought is to not have this very special recursion 00:10
and instead have simply a CREATEALL.
CREATEALL (CREATE+CREATE+CREATE) -> ...
so each CREATE only cares about the private atts
and CREATEALL cares about superclasses's layouts
that seems to make a lot more sense.
TimToady but in that case BUILDALL can probably do it. CREATE is just a preBUILDALL hook. 00:13
audreyt hm, but that means user-defined BUILDALL has to do a lot. 00:14
instead of only reordering the calls 00:15
00:24 weinig joined
audreyt cmarcelo: ivResponderInterface = class_interface base :: MethodTable Maybe is bogus 00:33
I'll take away the ::. :)
00:35 Aankhen`` joined
cmarcelo audreyt: ok. it is bogus indeed, but at some point the compiler (or my interpretation of it) was claiming this so I did it :o) 00:36
audreyt the compiler is claiming it because you didn't do the AnyResponder abstraction :)
tsktsk :)
(you should've seen the same claim for AnyClass also)
the idea is that the object has a AnyResponder 00:37
so it only existentially quantify the payload
the responder interface has nothing to do with the type of payload (ivInvocant)
so it cna be encapsulated safely
and that makes the explicit ::MethodTable go away
cmarcelo I see.. well I was lazy, when this classes appeared I was hoping to have a better solution than AnyMoose, so patched up this ::MT's.. 00:40
audreyt *nod* though AnyMoose is clearly the way to go now :)
cmarcelo rolling back to the layout discussion, when bless({x=>1}), the "layout" used will be Hash instead of P6opaque { ... }, and when using a Ruby object there'll be an ForeignObject { .. }, or these are different beasts? 00:44
audreyt that is correct
and the key conclusion of the discussion above is that you can inherit from a ruby object 00:45
and your CREATEALL (or, TimToady thinks BUILDALL last I checked) knows about it
so it delegates to the ruby bridge's CREAGE
CREATE
which creates a delegation
TimToady CREATE either has to allocate the slots or at least a handler slot for delegation 00:46
audreyt TimToady: but I think we need something to allocate private-slots only
TimToady I'd just like to avoid crawling the inheritance tree many times.
audreyt which is relevant if the class want One Specific Layout
but BUILD is usually undefined 00:47
and CREATE don't need to be fired if layouts are known to be compatible
i.e. if it's known in CREATEALL that things can be shared, it only calls one CREATE
00:47 hikozaemon joined
audreyt (which is another reason for the CREATEALL abstraction outside BUILDALL...) 00:47
but I'm not 100% sure yet myself. I think I'll just code it up and see how it smells like :) 00:48
TimToady I think CREATEALL is called "bless"
audreyt ...point. 00:49
TimToady maybe CREATE nullifies all the classes it creates, and then bless looks to see what's left
audreyt can we rename it to BLESS? :)
TimToady looks too much like BLAST! 00:50
audreyt (considering it's almost always autotriggered)
never mind... "bless" is fine
I jsut effectively murdered "new", we need at least one lowercase entry point :)
00:50 rodi joined 00:51 weinig is now known as weinig|food
TimToady and people may use new-less bless in P6 much like they do in P5. 00:52
at least for hashes
00:52 weinig|food is now known as weinig
audreyt I wonder if blessing arrays automagically works... 00:53
(ordering of attr is known.)
TimToady If you wonder, then it probably doesn't. :)
audreyt or if it's too tricky to be allowed this way
audreyt forsees painful bounds checking
never mind then :)
00:54 weinig is now known as weinig|food
cmarcelo audreyt: (/me reverse backlogging) using mono as runtime means some kind of PIL -> mono_byte_code? 00:56
audreyt cmarcelo: yes, but more importantly a runtime on mono that implements the MOP bridges 00:57
i.e. all our builtin objects (Scalar, Array, Class) must be encoded as native CLR objects
emeijer has suggested an encoding that works for Class
which was the hardest part
but the other ones need to be modeled as well
PIL itself is trivial.
00:59 aufrank joined 01:00 duend_ joined 01:02 mbradley is now known as mbradley|afk
cmarcelo audreyt: (Test.pm.yml is PIL serialized into yaml, right?).. I see.. but if they don't support something in CLR, e.g. MI, how's that possible? or the abstraction is lower? 01:02
audreyt the abstraction is lower. 01:03
the encoding is essentially to let metaclass control dispatch
not let VM handle it
if it's deemed to be incompatible with CLR's SI 01:04
we do it with a ProxyObjectDelegate
originally used for remoting
but can be used in-process as well
it lets an object control the complete dispatch for another object
which corresponds to AnyResponder
i.e. if we
instance ResponderInterface m (Invocant m)
then we also have that capability. 01:05
(am I making sense to you?)
cmarcelo actually, I got the workaround (let object control dispatch), but not the first option (letting meta control dispatch).. you mean their (CLR?) metaclass structure deal with our dispatch, if not possible, fallback to workaround..? 01:07
audreyt metaclass is just an object. :) 01:08
p6-metaclass controls the dispatch if needed 01:09
use native dispatch if possible
TimToady it's unlikely that their dispatcher will respond to "next METHOD"
audreyt correct, though if we are compiled closed/static
then we can know places where next METHOD will always not make sense.
01:10 lyokato joined
TimToady or if we use "hides" inheritance 01:10
audreyt in which case those classes can effectly become Mono structs
fortunately it seems the obj-control-dispatch, as long as that obj itself does not get proxied again, is very fast.
that amounts to not let people augment ::Class at runtime. 01:11
I'm willing to live with that when targetting mono :)
TimToady certainly these things should be negotiable
audreyt at some point something's gotta be nonaugmentable... 01:12
TimToady nonnegotiable is one of the things that's negotiable. :)
audreyt :D
01:12 eggzeck joined
TimToady use Perl#; 01:12
eggzeck Perl#? 01:13
audreyt TimToady: how is "submethod BUILD" different from "my method BUILD"?
oh, because you can't do $obj!*BUILD.
never mind, got it.
it neeeds to be made visible publicly but get dispatched privately. 01:14
TimToady nod
audreyt if someone defines "method BUILD" then very curious results will ensue...
namely derived chain calls it over and over again 01:15
TimToady one would presumably write it to derive class-specific info from somewhere else.
so it does different things when called on behalf of different classes
if there's a default Object::BUILD it works like that. 01:16
but probably autogenning BUILD for each class is more efficient
we could, however, make "method BUILD" barf by default. 01:17
lexical compiler warning, maybe 01:18
audreyt or metaclass object creation time warning. 01:19
TimToady mabye $.x vs $!x in submethod is a similar warning.
audreyt aka composition time
TimToady no warn "IKnowWhatIAmDoingDarnIt!";
audreyt seems to make more sense for a special-name-avoidance thing
*nod* 01:20
01:23 justatheory joined
audreyt cmarcelo: perl6 classes has this curious property: there's no separate classMethods and instanceMethods slots. 01:23
so I'm merging them... 01:24
01:27 Aankh|Clone joined
audreyt ditto roles. 01:28
01:28 Reinfield joined, Reinfield left
cmarcelo fine. but how are calls to accessors in classes are dealt? 01:28
re roles: how so? 01:29
audreyt calls to accessors always read from layout
it happens that the class prototype has a layout that fails on any access.
Point.x # fails 01:30
but that is only beause ::Point is allocated a NoLayout layout.
cmarcelo ok. and (obviously) merge them in roles too you mean... 01:32
audreyt yes but not bothering to do that tonight...
need to finish CREATE fist
first
but maybe I just checkin :)
adding perlcabal.org/~audreyt/tmp/for-cmarcelo-3 to Pugs. 01:33
lambdabot tinyurl.com/y2sndd
cmarcelo cool 01:35
audreyt it's in. 01:39
will give 6.4-compatbility a tiny try tomorrow 01:40
or you can do it
svnbot6 r14664 | audreyt++ | * Initial check in of the Haskell implementation of MO system.
r14664 | audreyt++ | Design: nothingmuch++
r14664 | audreyt++ | (with help from stevan++ and #moose)
r14664 | audreyt++ | Implementation: cmarcelo++
r14664 | audreyt++ | (with help from audrey++ and #perl6)
r14664 | audreyt++ | It's reasonably complete now, with attributes, accessors,
r14664 | audreyt++ | multiple inheritance, C3, full introspection, role composition
r14664 | audreyt++ | and conflict detection, autoboxing of native Haskell objects,
r14664 | audreyt++ | as well as the ability with all three of our Eval-monad runlevels
r14664 | audreyt++ | (Pure, STM, IO).
r14664 | audreyt++ | Too late tonight to hook it up with Pugs.AST and Pugs.Val now,
audreyt namely making existential records into positional
svnbot6 r14664 | audreyt++ | but stay tuned...
audreyt but if that alone doesn't do it, then we give up on 6.4 and mandate GHC 6.6.
cmarcelo++ nothingmuch++ # for making 6.28 possible 01:41
Pugs 6.28, that is :)
cmarcelo audreyt: ok, I could try that tomorrow night.. 01:42
01:44 Aankhen`` joined
audreyt cool. and I really shouuld blog about all those things (MP6, full bootstrap plan, native rules engine, native MO system...) 01:44
but sleep first. 01:45
see you tomorrow :) *wave*
cmarcelo yes blog! because "anarchy only works w/ communication" :)
s/s b/s, b/
audreyt or maybe _you_ write about it :)
you did 90% of the work anyway... 01:46
you have a blog bit, too :)
cmarcelo (you know more about how it integrates with the whole) :P
audreyt but you claimed you grokked the plan... :) 01:47
in any case, would be nice if you can writeup something, but don't feel too pressured.
I seriously need to sleep :) till tomorrow... *wave*
cmarcelo hehe I'll think about it... well, I have to go sleep too.. bye! & 01:48
01:48 cmarcelo left
audreyt nothingmuch: you have a blog bit too :) 01:48
& 01:49
01:51 avarab joined 02:00 weinig|food is now known as weinig, Southen joined 02:09 Pomin joined 02:17 brent[] joined
eviltwin_b hm. well, I definitely know where the filetest parse issue is coming from. but I have no idea what to do about it given the kluge contained therein 02:29
short version: "-" gets parsed before the filetest ops do, with @symbol@ so it consumes trailing @whiteSpace@. then there's a hack to check if it's actually a filetest op, due to something about precedence (which *may* in fact be an attempt to hack around "-" being seen before the filetest ops, but I'm not certain) 02:32
if it is, I can fix it easily enough, I think. but the comment on the filetest-op check mystifies me 02:33
02:38 awwaiid joined 02:47 weinig is now known as weinig|bbl 03:11 mdiep joined
eviltwin_b is trying the hack, with the previous kluge disabled, to see if it breaks parsing --- hopefuly there are tests in the tree for whatever triggered the old kluge 03:23
03:35 nipotaway is now known as nipotan 03:52 mako132_ joined 03:55 duend joined
eviltwin_b testing... 03:56
03:56 Lorn joined
eviltwin_b d'oh, forgot trailing whitespace. try, try again 03:59
04:41 mako132_ joined 04:58 nicodaemos joined
eviltwin_b hm, that was a pretty braindead thinko. but I think I might have it now. 05:01
05:09 Eidolos joined 05:10 duend joined 05:17 duend is now known as duend`sleeping 05:25 SubStack joined 05:29 lisppaste3 joined 05:55 beppu__ joined 05:58 baest joined 05:59 baest_ joined 06:14 baest joined 06:15 BooK joined 06:18 GabrielVieira joined 06:20 bennymack joined 06:53 marmic joined 06:59 drrho joined
svnbot6 r14665 | jesse++ | * "Differences from Perl6" didn't make sense in context. Relabeled the 07:06
r14665 | jesse++ | doc section to "Differences from Perl 5"
07:08 drrho joined 07:39 iblechbot joined 07:53 Alias__ joined 07:54 Alias__ joined 08:09 nusgnaf joined 08:11 justatheory joined 08:23 elmex joined 08:27 nusgnaf left 09:05 polettix joined 09:07 kanru joined 09:33 buetow joined 09:38 ruoso joined 09:43 andara joined 09:47 eggzeck joined 09:58 penk joined 10:08 zptao joined 10:21 kensanata joined 10:32 orafu joined 10:44 jabbot joined
svnbot6 r14666 | kudra++ | Summary from last week; commit failed on Sunday 10:46
r14667 | kudra++ | Placeholder 10:49
10:54 ruoso joined 10:56 todd- joined 11:00 Psyche^ joined 11:03 dmq joined 11:05 lumi_ joined 11:08 Psyche^ is now known as Patterner 11:19 xinming_ joined 12:00 elmex joined 12:02 lumi_ is now known as lumi 12:04 fglock joined
svnbot6 r14668 | fglock++ | * MP6 12:04
r14668 | fglock++ | - implemented MP6::G::Regex::rule()
r14668 | fglock++ | - /<%hash>/
r14668 | fglock++ | - /const|const/
r14668 | fglock++ | - /<subrule>/
r14668 | fglock++ | - /<'xyz'>/
r14668 | fglock++ | - /\n/
r14668 | fglock++ | * v6 - added more runtime methods
fglock the Token -> MP6 compiler mostly works now 12:07
12:14 chris2 joined 12:22 beBOOm is now known as boo_ 12:24 Psyche^ joined
fglock MP6 now needs a Runtime 12:26
it needs Match, infix:<+>, substr, m:P5//, <ws>, <alpha>, <digit> - I guess that's all 12:31
12:32 Psyche^ is now known as Patterner
dmq neat 12:34
gaal nothingmuch: ping 12:35
12:40 mako132_ joined
xinming_ fglock: Do you mean MP6 is almost finished? >_< 12:43
12:43 xinming_ is now known as xinming 12:53 Limbic_Region joined
Limbic_Region use.perl.org/~acme/journal/31531 12:55
lambdabot Title: Journal of acme (189)
Limbic_Region "Of course, you really want to see the opcodes - hey, it's real bytecode. Now how long will it be until Pugs targets Tamarin directly?"
12:55 iblechbot joined 13:03 rhizo joined 13:04 fglock joined
fglock xinming_: no, it will still take a few days to finish 13:06
it has been hard to debug it 13:07
Limbic_Region: re Tamarin - just write an emitter :) 13:08
Limbic_Region fglock - yeah sure, I will do that on my lunch break 13:10
13:15 rodi joined
svnbot6 r14669 | fglock++ | * MP6 - '/' can end a regex 13:15
fglock Limbic_Region: v6-MiniPerl6/lib/MiniPerl6/Emitter.pm - you can save it to Emitter/Tamarin.pm 13:16
gaal moose. has the sv? repo moved yet? 13:18
fglock gaal: I think not 13:19
13:40 weinig joined
svnbot6 r14670 | fglock++ | * MP6 - start integrating the Regex subcompiler into the main grammar 13:43
audreyt gaal: scheduled in ~12hr from now. if I got too tired watching FIFA beach soccer world cup then 36hr from now 13:52
but I think 12hr is correct.
oh hey! the JS2 VM promised by brendan is finally here! 13:58
gaal okay, thanks. I've been too out of things lately :(
audreyt we can finally target JS without massive pain and performance penalty! 13:59
mozilla++
13:59 stevan joined
audreyt tamarin++ 13:59
a JITing JS2 VM 14:00
14:07 bonesss joined 14:11 c6rbon joined 14:20 mbradley|afk is now known as mbradley 14:21 [particle] joined
nnunley Wow. JS2 is out? 14:27
14:37 fglock joined 14:45 vel joined 14:49 chris2 joined, penk joined 14:52 ludan joined 14:53 rlb3 joined 14:54 Debolaz2 is now known as Debolaz 14:55 vel joined 14:57 duend`sleeping is now known as duend 14:59 elmex joined 15:05 hexmode joined 15:14 cjeris joined 15:25 penk joined 15:32 penk joined
svnbot6 r14671 | fglock++ | MP6 - intermediate regex compilation finished... 15:35
15:44 autark_ joined 15:52 hexmode joined 15:56 fglock joined
GabrielVieira fglock go sleep a little, please :P 15:57
fglock Gabriel: how was the trip back?
is openfoundry down again? 15:58
15:58 avarab is now known as avar
fglock audreyt: MP6 can now compile token{...} - only needs a few tweaks (but I can't commit) 15:59
clkao why can't you commit?
fglock could be bootstrapped today
clkao: I can't ping openfoundry 16:00
clkao: what is the svk command to do a local ci ?
clkao fglock: svk cp //mirror/pugs //local/pugs; svk sw //local/pugs 16:01
then you can just commit
(assuming //mirror/pugs is your current checkout)
GabrielVieira wow 16:04
let me read this
fglock was really nice... and fast. I had no problem to come back :) 16:05
16:05 buu joined 16:08 buubot joined, GnuVince joined 16:18 kanru joined
nothingmuch autark_: i don't really uinderstand larry's rulings from your line (haven't backlogged yet) 16:31
err
audreyt: bah 16:32
why is the tab completion not in lexical order =/
anymoose
as far as I'm concerned Class/Role/* etc are entirely flexible
that is Perl 6 should come with some default ones
that adhere to @Larry's design
if possible we refactor them so that this base set of metaclasses is made out of reusable components 16:33
(more on that in the coming wekks, i guess)
fwiw what I sketched up was a bit of moose and a lot of perl 6 but by no means anything definite 16:34
16:36 polettix joined 16:47 Limbic_Region joined
audreyt sure 17:07
17:33 < nothingmuch> if possible we refactor them so that this base set of metaclasses is made out of reusable components
and that's a good plan.
17:11 andara left
fglock audreyt: MP6 is nearly ready to parse itself 17:12
17:17 bpphillips joined 17:26 thepler joined 17:28 elmex joined
audreyt fglock++ # woot 17:28
openfoundry seems to work here
fglock I can't ping it - is it 140.109.17.109 ? 17:31
TimToady yes 17:32
Limbic_Region pinging isn't always a good indication of availability
ICMP may be disabled where as TCP port 80 or 443 may be enabled 17:33
fglock trying again
17:34 TSa joined, rodi joined
fglock no luck 17:37
audreyt :/
Gothmog_ What does nmap say?
audreyt ok, that increased priority of the move
17:38 autark joined
clkao audreyt: let's do it 17:38
do you have uptodate mirror?
dump the revs with pullyu -r
audreyt I have uptodate mirror 17:39
do I need to redump everything?
I thought you had something on feather that you can just sync. 17:40
clkao no
sync from where?
audreyt: you need to dump the revs i don't have on feather 17:41
GnuVince Good day everyone 17:42
clkao audreyt: pullyu -r14582@:HEAD ~/.svk/local /mirror/pugs
fglock brb 17:43
TimToady is there some way to make openfoundry ro so we don't get conflicting checkins? 17:44
audreyt Can't locate SVN/Dump.pm
unlisted dependency, fun
TimToady: yes, openfoundry has a knob
but the webui is currenly down
so I can't flip it
otoh, that also means commits can't ge through
so it's essentially RO ;)
TimToady and middle of the night there... 17:45
so unlikely to change
if history is any guide...
clkao audreyt: also you need 0.02 svn::dump. at the time it wasn't released
[particle] what happens, does the gerbil fall asleep and stop running the wheel that powers the net connection? 17:46
clkao you can poison the svn.openfoundry.org dns though
17:46 hexmode` joined
TimToady that sounds antisocial 17:46
clkao audreyt: wait a sec, 0.02 has incompat api 17:47
change
audreyt yeah, that kills all other openfoundry projects
clkao: mmm software! point me a tarball please.
make me one with everything (to install)
17:47 hexmode left
clkao the one on cpan. but i am going to test this shit 17:47
17:47 hexmode` left
clkao audreyt: ENOBRAINPOWER. get 0.01 and apply my patch: rt.cpan.org/Public/Bug/Display.html?id=22429 17:53
i hate software
audreyt thanks for your instruction :)
lambdabot Title: #22429: Allowing reading from fh, tinyurl.com/vb9vu
obra loves how ENOBRAINPOWER matches ~[6~[6~[6~[6~his irc nick 17:54
fglock :)
audreyt En Obra In Power 17:55
TimToady is there an easy way to use "svn lock" to lock everything, or do you have to feed it all the names?
or would that fail right now too?
audreyt lock locks subtree.
but svn.openfoundry.org uses a subrequest to the website for auth 17:56
so when website is down, auth always fail.
TimToady *nod*
clkao orafu: hahaaa
obra: haha
audreyt welcome our new obra-in-power overlords 17:57
17:57 ludan joined
audreyt clkao: is 1.4.0 binding fine for svk on feather 17:57
or would you require some sort of trunk svn?
obra audreyt: that would be the commitbit 17:58
fglock [OT] any good reference on connecting to Windows shares from OSX?
audreyt fglock: simply click on finder, then press apple-k
clkao audreyt: you want 1.4.2
audreyt or on the menu, "go" "connect to server"
clkao audreyt: well, depends on what you want to do on feather
audreyt clkao: commitbit + webdav 17:59
clkao it doesn't need 1.4.2 if you are just being the master. pushmi would require 1.4.2
as slave
audreyt ok. curiously the latest release is 1.4.0
are they going to skip 1.4.1 or something?
clkao audreyt: also if you have 1.4.2 and use svkmirror-pipeline branch, mirroring is superfast
yes
1.4.2 is rolled
it's in the collab download directory
not announced yet
18:00 buetow joined
audreyt clkao: 1.4.2 on server+client? 18:01
or just client?
clkao client
audreyt oh. so 1.4.0 on server is sufficient
for the time being. ok 18:02
clkao yep
audreyt running pullyu... 18:03
18:04 loumz joined
audreyt clkao: ~audreyt/worktmp/delta-moose 18:07
now what?
commitbit first, or moddav_svn?
TimToady um, openfoundry seems to have come up 18:08
I got the admin page, anyway...
audreyt really? I didn't 18:09
clkao audreyt: dav_svn first
audreyt TimToady: if you got the admin page please click on the "version control" tab and switch to read-only
clkao audreyt: ~clkao/pugs.repo is uptodate
technology++ 18:10
audreyt clkao: ok. let me put it to var
clkao computer: ę‚Øéœ€č¦äø€å€‹ä¼‘ęÆ
audreyt gives clkao a ^C
TimToady I do not see such a button 18:11
audreyt clkao: er feather is running apache 1.3 18:12
audreyt ponders what to do
TimToady perhaps a mere admin is not so privileged...
maybe I can lock it
18:12 justatheory joined
audreyt please lock it then 18:14
clkao that means you didn't compile the 1.4 with dav_svn anymoose
audreyt there's apache2 18:15
just not runnning
clkao ah
audreyt maybe I can have ap1 proxy to ap2 on another port
sec
audreyt remember she has done this ~5 times on openfoundry
clkao what's the apache1 doing? 18:16
audreyt cgi-irc
squirrelmail
jrockway svnserve?
clkao modperl invovled?
jrockway that's how I cope with svn+apache1
audreyt svnserve is not linkable
we despreately want http://
jrockway true, true
SVN::Web, maybe? 18:17
fglock audreyt: apple-k worked, but i get ".. one or more required items cannot be found Error code -43" - is this related to the sudo problem?
audreyt fglock: no, I fixed the sudo problem
I don't know :/
fglock: try mounting FTP?
fglock ok 18:18
audreyt: did mk find you? 18:19
audreyt fglock: she mailed me and gave me time+place to meet tomorrow 18:20
fglock nice
audreyt clkao: in download area I don't see 1.4.2 tarball 18:22
oh nvm
clkao subversion.tigris.org/downloads/
audreyt found www.red-bean.com/maxb/svnrm/1.4.2/s....2.tar.bz2 18:23
Limbic_Region two smart chicks in one place - every geek's fantasy
lambdabot tinyurl.com/wd9m4
TimToady okay, I think I switch rt.openfoundry.org to readonly by telling it to use svn.openfoundry.org/perl6 as its remote RO repo
It does seem to prevent checkins, anyway... 18:24
clkao TimToady++
TimToady later we can make it remote to feather
audreyt very nice
elmex i've problems with pugs, Makefile.PL doesn't find 'ghc' but i've installed ghc-6.5, but it isn't 'ghc' in PATH. can i tell the pugs Makefile.PL somehow to use 'ghc-6.5' ? 18:25
clkao TimToady: the last is r14671 ?
TimToady couldn't get it to lock the top dir though...
yes, r14671
Limbic_Region elmex - I use ghc-6.5 without a problem
audreyt elmex: export GHC=/path/to/moose
clkao audreyt: tell openfoundry they need to install pushmi
TimToady whenever I tried "svn lock ." in pugs it said it couldn't lock my home dir. seems like a bug
audreyt clkao: I don't know anytihing about pushmi, so how about you tell it.
Limbic_Region elmex - the executable itself isn't called ghc.exe ? 18:26
clkao audreyt: i will put it on bps-public after dinner
elmex Limbic_Region: i'm on GNU/Linux (on a amd64)
Limbic_Region or just ghc on a *nix platform
elmex Limbic_Region: .exe file endings are uncommon on that platform
Limbic_Region my question is - is it a PATH issue or is it executable name issue
if it is a PATH issue, then it should be simple to solve 18:27
if ghc the executable doesn't exist it may be more of a problem
audreyt Limbic_Region: ghc-6.5 on win32 is still called ghc.exe
elmex Limbic_Region: i guess the issue is just that 'ghc-6.5' ne 'ghc'
audreyt if you rename ghc-6.5.exe on win32 then it will still fail to probe
Limbic_Region audreyt - I know, and it works great - which is why I am confused
audreyt - I know that too
audreyt - that's why I was trying to acertain if elmex had changed the executable name 18:28
audreyt Repository Switch In Process - Commits Temporarily Disabled | run.pugscode.org | spec.pugscode.org | paste: sial.org/pbot/perl6 | pugs.blogs.com 18:28
audreyt ah. ok 18:28
elmex ah :)
Limbic_Region elmex - as part of a path name, it shouldn't matter
as part of the executable name it will
ashleyb fglock: is miniP6 going to obsolete P::C::Rule? I've found some more bugs, are you interested in tests? 18:29
audreyt nothing will ever obsolete tests :) 18:30
ashleyb yes but the tests are p5 tests.
audreyt PCR will remain a perl5 module.
emitted-to-p5 module, that is :) 18:31
i.e. interface should stay
ashleyb true, but is development going to be discontinued?
ok
elmex where can i read about applying the tests in t/ to a dofferent perl6 implementation than pugs?
audreyt elmex: "make test-perl5" does that
elmex ah, ok
thanks
audreyt clkao: did you take care to make uuid agree with upstream? 18:32
in pugs.repo that is
clkao yes 18:33
pullyu does that automagically
it's design to reproduce the original repository
fglock ashleyb: yes, tests are good!
ashleyb cool.. 18:34
fglock PCR is only going to be obsoleted if MP6 works really well :)
TimToady so far it's just a bunch of MP promises... <ducks> 18:35
audreyt indeed, before te emitter+runtime really works, it amounts to a bunch of AST floating above the cloud :) 18:37
[particle] is it MP-complete yet? 18:38
TimToady no, but it's looking pretty 6y. 18:39
Limbic_Region [particle] - no, just MP-hard
TimToady Any ETA on the new Rushing Featheration? 18:40
audreyt the repo is converted. I'm setting up dav 18:41
read-only access expected to be restored in <30min
then it's setting up commitbit
and resend all the invications 18:42
18:42 todd- joined
audreyt that can take another hour or so. 18:42
TimToady "I'm sorry Dav, I can't allow you to do that..."
audreyt fortunately it's not even 5pm here
fglock audreyt: re error -43 - I was typing the wrong share name :P everything i fine now
TimToady k, guess I'll go take a cold shower...
fglock is 18:43
18:51 vel joined 18:58 vel joined
audreyt dav installed. proxy installed. 19:02
just a sec...
19:03 eggzeck joined
GnuVince audreyt: I want to buy a few items from cafepress.com/pugscode. Does the money go to you? 19:07
audreyt GnuVince: they do 19:09
pasteling "fglock" at 200.17.89.88 pasted "intermediate compilation of token{abc} to MP6" (4 lines, 335B) at sial.org/pbot/20952 19:10
GnuVince audreyt: I'll throw in an extra mug then :) 19:11
audreyt :D 19:12
19:12 fglock joined
audreyt finds debian surprisingly sane despite never havingu used it 19:12
clkao most OS still alive are sane to some extend... 19:13
audreyt *cough* AIX
19:13 rindolf joined
rindolf Hi all. 19:14
What, no committs????!
clkao how about you try committing
audreyt rindolf: relocating to a server that won't shut down two days per week
rindolf clkao: I don't have anything to commit.
audreyt clkao: should I simply Include the .conf written by commitbit?
clkao yes
rindolf audreyt: where is this server going to be located?
fglock audreyt: see nopaste - now parsing "method"; after that I'd like some help to make it sane :)
clkao audreyt: it should have the authz etc all set for you 19:15
ingy AIX was dead 15 years ago
clkao ingy!
ingy hi clkao! 19:16
audreyt ingy: that makes it a long-lived undead
clkao how are you!
audreyt aka -- lich!
ungratefaul undead operating systems...
ingy is surprised at being greeted with excitement from cl
clkao: I am pretty well. where are you today?
clkao london, baby 19:17
ingy going to Taiwan any time soon?
clkao next month
ingy how long?
clkao dunno 19:18
ingy I'd like to flatshare some time with someone in .tw for a month or so...
pasteling "ashleyb" at 66.201.51.66 pasted "PCR tests (for fglock) -- variations on a theme: more hanging tests (and some similar ones that pass)..." (42 lines, 929B) at sial.org/pbot/20953 19:19
ingy but that thought is vague and noncommital at this point
audreyt clkao: commitbit sets up both authz and basic auth.
correct?
clkao audreyt: yep 19:20
fglock ashleyb: thanks! 19:22
rindolf Hi ingy 19:23
audreyt trying svk mi --relocate 19:24
clkao where to? 19:26
19:28 mosca joined 19:32 GabrielVieira joined
TimToady 'course, it's possible the jinx will now be transferred from openfoundry to feather... 19:33
audreyt clkao: weird, PROPFIND still doesn't seem to be proxied 19:34
GET etc is fine
clkao iirc you need to speicfy what methods you want to proxy 19:35
it's a bit painful
how about put apache2 at 80 and proxy the apache1 stuff
audreyt clkao: can you find the relevant conf a bit? 19:36
audreyt is googling but doesn't seem to yet turn up things
clkao h6w ab64t 4se -er3ba3 19:37
obra to proxypass all methods from an apache1 to apache2?
clkao how about use perlbal ;p
audreyt obra: yes
obra hang on.
audreyt clkao: stop introducing new complexities please :)
clkao it's already included in pushmi :P 19:38
audreyt: trailing slash for your Proxy command? 19:39
obra ProxyPass /svk localhost:8800/svk
ProxyPassReverse /svk localhost:8800/svk
actually works for me
I don't have special Limit or LimitExcept lines 19:40
audreyt a sec.
19:42 fglock joined
audreyt and in 8800 it's <Location /svk> 19:42
not <Location /svk/> ?
obra loking 19:43
fglock home &
obra Location /svk 19:44
as generated by commitbit
(this is the commitbit-generated httpd.conf fragment)
audreyt it is the trailing slash.
clkao++ # entirely correct.
changing /pugs/ to /pugs solved it.
obra commitbit needs more magic to detect and fix slash issues.
audreyt it's actually Proxy layer :) 19:45
obra but first we'd need tuo understand what's right ;)
audreyt $ svk mi --relocate svn.pugscode.org/pugs/ /pugs/trunk
lambdabot Title: Revision 14671: /
audreyt Mirror relocated.
woot!
clkao :D
audreyt New Repository - svn.pugscode.org/pugs/ - Commit Bit Will Arrive By Mail Shortly | run.pugscdoe.org | spec.pugscode.org | paste: sial.org/pbot/perl6 | pugs.blogs.com 19:46
obra Retrieving log information from 10641 to 14671 19:46
clkao obra: you using pipeline?
obra clkao: nein. still on 1.40
audreyt TimToady: as it seems I can't access openfoundry, can you relocate its mirror when you next get cycle?
clkao i would be very surprised if openfoundry can be turned from master to a ro mirror of something 19:47
[particle] how do i change my wc to point to the new repo? (svn)
kolibrie audreyt: topic typo: run.pugscdoe.org
clkao svn sw --relocate OLD svn.pugscode.org/pugs/
lambdabot Title: Revision 14671: /
audreyt clkao: I hacked in that feature thank you very much :)
TimToady done
audreyt (to RT-Foundry)
[particle] clkao++ 19:48
19:48 electrogeek joined
obra audreyt: evil 19:48
audreyt obra: part of original spec 19:49
obra aw. no code.pugscode.org
audreyt it's commitbit.
TimToady hmm, now going through openfoundry I get (for svn up):
svn: REPORT request failed on '/pugs/!svn/vcc/default'
svn: No such revision 14671
audreyt not code.
obra *nod*
audreyt walking advertisemnt :)
obra Thanks :)
we need a logo
audreyt TimToady: wait a while for it to resync 19:50
integral yay, feather's much faster for me :)
audreyt glad to hear that! 19:51
now setting up commitbit.pugscode.org...
clkao audreyt: you might want to turn on compression as well 19:53
audreyt svn.pugscode.org/pugs/ - Commit Bit Will Arrive By Mail Shortly | run.pugscode.org spec.pugscode.org | paste: sial.org/pbot/perl6 | pugs.blogs.com 19:53
audreyt clkao: how do I do that? 19:53
obra enable mod_deflate in apache2 19:54
audreyt working on it
dmq timtoady: did you get a chance to consider the regex stuff I asked about yet? 19:56
audreyt will SetInputFilter DEFLATE also help?
GnuVince S03 mentions that the range operator doesn't do backward ranges. What if there was an operator to do it? I suggest inverting the dots to signify this intention ;-) 19:58
audreyt what a fabolous idea 19:59
so they will appear at top of the characte box
instead of at the bottom baseline
GnuVince ~(1..3) eq "1 2 3"; ~(3ōæ½xA81) eq "3 2 1" 20:00
Khisanth stabs GnuVince 20:01
that looks like two specks of dust on the screen
obra audreyt: I haven't tried the SetInputFilter method 20:03
GnuVince Yeah, they're not very visible. Not sure what the accent name is in english, but that's a "trōæ½xE9ma" (french), the double-dots you put in the word Noōæ½xEBl
Khisanth: and fear not, I wasn't actually suggesting it. 20:04
audreyt gzip is on. 20:06
20:07 bernhard joined
eviltwin_b "double-dot diacritical". ("diaresis" technically refers to the meaning in French and English, not the symbol) 20:08
GnuVince eviltwin_b: Wow, thanks. That's quite a mouthful. 20:09
audreyt clkao: pipeline branch:
RA layer request failed: REPORT request failed on 'svn.pugscode.org/pugs': REPORT of 'svn.pugscode.org/pugs': 400 Bad Request (svn.pugscode.org) at /opt/local/lib/perl5/site_perl/5.8.8/SVK/Mirror/Backend/SVNRa.pm line 241.
hm wait. maybe my conf error 20:10
eviltwin_b as for operators: keep .., as a symonym for .>. , then .<. can be the reverse 20:13
? 20:14
audreyt visually confusing with .>>. though.
eviltwin_b (can't put the arrpw on the outside, that looks too much like hyper ops
audreyt $x.>>.sort
is hyper
eviltwin_b ah
I missed that opne, thought all hyper ops were on the outside
GnuVince reverse 1..3 sounds perfrectly reasonable. 20:15
audreyt .oO
Oo.
eviltwin_b oh wit, it is, that's the dpt operator
my, what a tangled web we weave...
eviltwin_b can't type
"oh wait, that's the dot operator" 20:16
audreyt hm, I wonder if we can make a cert for feather that also lists svn.pugscode.org.
until that's solved, an additional warning may appear 20:17
but I've just set up svn.pugscode.org/pugs/
for people in .cn have reported that https is easier to get through.
clkao or there should be a cn site powered by pushmi.. 20:18
audreyt pipelining appears to work.
not as in speed, as in not crashing.
clkao not feeling it faster than raw sync?
audreyt for 0rev 20:19
they seem to be the same
;)
maybe I should try a full sync to test.
clkao or run two syncs side by side
audreyt https and http should be same in speed right?
(roughly) 20:20
(with pipelining)
or is http:// still faster?
clkao i don't know
audreyt ok, I'll bench.
clkao you can do live test ;)
audreyt considering Basic Auth, I feel more comfortable with https
clkao right 20:21
what was openfoundry repository using again?
20:21 weinig is now known as weinig|bbl
audreyt Basic, but then it also has https 20:21
so I was always using https
clkao it did?
audreyt yes, by request of agentzh relayed by me
clkao oh well, anyone who wants to temper the pugs repository could just come here and ask for commitbit ;) 20:22
audreyt exactly ;)
it's not as if we have something to hide ;)
clkao i guess it's still different from "tempering under audrety's name"
audreyt I'm fine with people committing as audrety
committing as audreyt however is more trouble :) 20:23
clkao i haven't got commitbit mail.. .
audreyt pipelining appears to be faster by about 40%
but that's just avg'ing 10revs of data before I ^C'ed it 20:24
no you havn't because I was setting up gzip and https instead of commitbit! :)
audreyt goes back setting up commitbit.
TimToady just changed openfoundry to use https, but it still gives same REPORT error 20:25
clkao audreyt: yayy
TimToady what exactly is doing this resyncing that I'm supposed to wait for? 20:26
and did I just tell it to restart by changing to https? :/ 20:28
audreyt you... did
and because it blocks on auth
it will never happen :D
(credential requires a "p" from tty) 20:29
TimToady should it use 's' or not?
audreyt please change back to http
not 's'
and pray that it will resync instead of blocking
audreyt crosses her fingers
TimToady k
audreyt whew it's resyncing. 20:30
so just wait for a couple hours
:)
TimToady voting &
audreyt meanwhile goes back setting up commitbit For Real
audreyt praises BestPracticalSolutions for software that (mostly) Just Works 20:31
20:39 b_jonas joined, larsen_ joined, b_jonas left
audreyt clkao: from Rio here, pipelining is exactly 2x the speed than non- 20:40
taking 50%~60% of the time
clkao: great work :)
(if you deduct the initial logsyncing which takes longer, it seems to be exactly 2x)
20:41 duend is now known as duend`school
clkao hurrah 20:42
I ROCK 20:43
audreyt clkao++ clkao++ 20:44
does that make DAV bearable? :)
clkao ya
a little bit
20:48 duend`school is now known as duend
duend exit 20:48
20:48 duend left
dmq could somebody tell me what 'CY'=~ms/ [ [ A :: | C :: | D :: ] X | [ C :: | D ::] Y ] / does? 20:52
20:52 DHGE joined
dmq eval 'CY'=~ms/ [ [ A :: | C :: | D :: ] X | [ C :: | D ::] Y ] / 20:53
audreyt what's ms? 20:54
?eval 'CY' ~~ m/ [ [ A :: | C :: | D :: ] X | [ C :: | D ::] Y ] /
oh, match with sigspace.
ERROR - Error sending mail: error when closing pipe to /usr/sbin/sendmail: No child processes 21:00
bother...
obra ouch 21:01
dmq well, i dont really mind what kind of match it is im just wonder if it matches. 21:03
or rather if its supposed to match.
21:03 boo_ joined
dmq does the '::' fail the whole match? 21:03
[particle] no, it fails the group 21:04
::: fails the whole match
21:04 hexmode joined
dmq so it fails the inner group and then tries the second? 21:04
[particle] yes 21:05
dmq so when you say ::: fails the whole match, what would happen if the input string was CYCX and the pattern was [ [ A :: | C ::: | D :: ] X | [ C :: | D ::] Y ] 21:07
(note the use of a ::: )
would it match the 'CX'?
21:08 electrogeek is now known as electrogeek_brb
[particle] yes, i believe so 21:08
audreyt bother...
er, sorry, ww :)
(but that was accurate as well.)
dmq hrm, im trying to figure out if ::: is what im calling '(*CUT)' 21:11
sorry for the strange questions
[particle] np, hang on and i'll get pmichaud in here 21:12
pmichaud_ here
oh, input string of CYCX on that pattern fails
dmq fails?
so its not what im calling (*CUT). shoot. 21:13
so then i dont see the difference between ::: and <commit>
pmichaud_ ::: fails a rule, <commit> fails a match
[particle] oh, it'll match the C, then fail on the X because the ::: causes the whole match to fail on backtracking
pmichaud_ the difference is when subrules are involved
a rule containing ::: will fail that particular rule (if it was invokved as a subrule), but the outer match will continue 21:14
a rule containing <commit> will cause the outer match to fail also
dmq ok, so when no subrules are involved <commit> and ::: are equivelent?
pmichaud_ effectively, yes.
dmq hrm. ok.
a single ':' seems like a post op way to have a (?>....) wrapped around the previous atom. 21:16
is that about right?
pmichaud_ pretty much. Syntactically it's much cleaner
PGE implements ':' as a postfix quantifier 21:17
dmq ok, so perl5 is doing that with possessive quantifiers, but its basically the same thing.
so iiuir X+: in perl6 is the same as X++ in perl5.10
since X++ is the same as (?>X+) 21:18
pmichaud_ I'd have to look up the X++ syntax (haven't studied perl 5.10 much)
21:18 prefiks joined
dmq its new. 21:18
pmichaud_ but yes, X+: is (?>X+)
[particle] is ++ friedl's?
if so, yes 21:19
dmq ok, so then we cant do '::', but it looks like ::: is what im calling (*COMMIT)
++ is the notation friedl recommended. It originates in a java package but i forget which.
pmichaud_ I like the perl6 notation much better. Of course, that doesn't really work for p5, because : isn't meta in p5 21:20
dmq no. :-)
im going to have to figure out a name for '::'
[particle] (*CUTGROUP) ? 21:21
pmichaud_ PGE uses '::' => 'cut group'
':::' => 'cut rule'
'<commit>' => 'commit'
<commit> is also known as "cut match" in PGE 21:22
dmq (*CUT) is different in its current existance.
it tells the engine to skip the already matched characters when looking for a valid match. 21:23
pmichaud_ like <cut> ?
dmq my understanding is cut is a little different.
your cut i mean.
[particle] what happens if you backtrack past (*CUT)? 21:24
dmq for instance 'aaaaabaaaab'=~/a+b(*CUT)(*FAIL)/ would first try at pos 0, match up to the 'b' fail, and start again at the 'a' following the first 'b'.
pmichaud_ hmmm 21:25
dmq particle: it fails the match at the current startpoint, and advances the startpoint to at earliest the cursor point where the cut was encounterd.
pmichaud_ it's pretty close to a p6 cut, with the exception that a p6 cut also fails the match
[particle] ok
p6cut is destructive 21:26
audreyt clkao: htdigest support possible?
21:28 justatheory joined
dmq think about what 'aaaaabaaaab'=~/a+bc/ would do if the optimiser werent smart enough to look for the 'c' 21:30
clkao audreyt: you ask for a lot ;)
dmq like for instance: 'aaaaabaaaab'=~/a+b[XYZc]/ 21:31
pmichaud_ PGE doesn't have that optimization yet, but it will
PGE will be smart enough to know that /a+b[XYZc]/ is the same as /a+:b[XYZc]/
dmq Cool! ( PCRE already is. And its on my todo list for perl5. :-) 21:32
audreyt clkao: you can say "no" :)
dmq but anyway, forget that point.
consider how much work happens with 'aaaaabaaaab'=~/a+b[XYZc]/
clkao audreyt: not too soon. but you have a commitbit
dmq wheras 'aaaaabaaaab'=~/a+b(*CUT)[XYZc]/ would do a fraction
clkao i thinik it's a matter httpd.conf 21:33
audreyt ok. once the critical injection attack is fixed I'll happily do that
dmq do you see what im saying pmichaud? 21:34
pmichaud_ oh, yes, I see 21:35
dmq its like an optimisation hint. 21:36
pmichaud_ in p6 rules we would do / [ a+b :: ] <[XYZc]> / 21:37
dmq ive actually got perl5 working so that you can set the point independently of the (*CUT).
pmichaud_ but hmmm, that still doesn't get the same behavior
I'll have to think about that one a bit
dmq incidentally my (*COMMIT) is just a special case of my (*CUT), on failure it sets the "cutpoint" to the end of the string. 21:38
audreyt clkao: you got spam. can you try commit to pugs? 21:39
21:40 bpphillips left, justatheory_ joined
dmq btw, all of this is in blead perl if you want to play with it. 21:40
pmichaud_ that would be cool 21:42
tnx
afk & # making dinner
dmq cool, me too. 21:43
audreyt svnbot now checks every 30 seconds 21:45
since it's a file:/// check :)
21:45 svnbot6 joined
[particle] that's a nice improvement :) 21:46
...and side effect
audreyt :D
21:46 weinig|bbl is now known as weinig 21:47 eggzeck joined 21:49 frederico joined
audreyt clkao: so. to sync back into svn.perl.org 21:50
and to svn.openfoundry.org
pushmi is cleanest right?
compared to svnsync, svk smerge
21:52 Debolaz2 joined
clkao audreyt: pushmi has to be run on the slave site 21:55
the good thing is that it also takes commit
so it's seamless
audreyt oh hm.
clkao to clients
audreyt so we need to talk to Robrt? 21:56
to make svn.perl.org a pushmi mirror that is
or one of the new boxes
commitbit.pugscode.org is almost ready. 21:58
pending a single vendor fix :)
21:59 jferrero joined
audreyt oh it's in. 21:59
obra++
[smash] hi all 22:09
smash_ anyone here working on pugs/trunk/v6/v6-MiniPerl6/ ? 22:10
audreyt smash_: me, and I suspect fglock in a couple hours 22:11
smash_ nice 22:12
any docs/papers/examples i could read about it ?
(looking at the code now) 22:13
audreyt smash_: there is pugs.blogs.com/photos/visiolization...strap.html
lambdabot Title: Visiolization: Mp6bootstrap, tinyurl.com/yxb6ux
audreyt there is also t/
not much beyond those for now
a HACKING file would be nice 22:14
smash_ cool.. let me take a look at those
audreyt but I'm swamped in apache configuration land, so please bear with me :)
smash_ heh
ahh.. apache configuration land ? not that bad
(could be worse) 22:15
obra . o O { Could be AOLServer}
audreyt apache 2.2 is actally good. 22:16
smash_ need any help with apaches .conf ?
smash_ still sticks to 1.3
pasteling "ashleyb" at 66.201.51.66 pasted "(for fglock) -- here are some more detailed tests for that hang problem." (55 lines, 1.4K) at sial.org/pbot/20958
ashleyb @tell fglock here are some more detailed tests for that hang problem. --- sial.org/pbot/20958 22:17
lambdabot Consider it noted.
22:17 rlb3 joined
smash_ ahh.. haskell function declaration.. neat 22:28
22:30 rodi joined, mAdzzzzzz joined
audreyt commitbit.pugscode.org/ is up. 22:31
lambdabot Title: Welcome to CommitBit for our projects
jrockway hmm, the links seem broken 22:32
they link to perlcabal.org
which is 404-ing
22:32 Limbic_Region joined
audreyt really? 22:33
try force reload
smash_ it looks fine to me 22:35
jrockway weirdness
now it works, but it's at svn.pugscode.com:9999
(it wasn't before)
audreyt nod. it was wrong for a minute. 22:36
[particle] i've seen the same behavior
audreyt in any case everything should work now. 22:37
scripting the commit-bit-resender with pugs...
jrockway heh, does jifty run on top of pugs yet? 22:38
audreyt if you consider
use perl5:Jifty;
then yes ;) but no, we havn't ported Jifty to run on v6.pm.
svnbot6 r14673 | audreyt++ | * Massive revisionism: Change all files in docs/ to mention svn.pugscode.org 22:43
r14673 | audreyt++ | not svn.openfoundry.org.
audreyt yay, svnbot works. 22:44
[particle]: you got a minute?
or anyone who want to volunteer for commitbit testing before I spam everybody? :)
[particle] ok 22:45
spam me! spam me!
audreyt k. checkmail
then do a trivial commit 22:46
then goto commitbit.pugscode.org and invite the next victim^Wvolunteer
so we can make sure the entire chain works
chain, as in chain mail
TimToady the page doesn't list invites sent? actually, lists them as fait accompli...
audreyt which provides pretty good AC
[particle] i'm logged in
audreyt [particle]: try commit 22:47
[particle] hrmm
audreyt TimToady: correct
[particle] it said i was logged in
then i clicked 'Preferences' and it sent me to login screen
audreyt then you should be able to set password or something
that link is only good for setting pass, I think...
or is the link broken?
/msg me the link you got from mail? 22:48
[particle] well, now that i'm *really* logged in, there is no link
for preferences
audreyt it should be svn.pugscode.org:9999/prefs
lambdabot Title: Login
[particle] that points me to login 22:50
lemme log out
audreyt ok
[particle] reload is slow
smash_ it should have something more after prefs
logout & login 22:51
(or clena local cache)
using firefox ?
[particle] yes ff2
smash_ sometimes firefox can be a pain
[particle] when i'm logged in, i have Home | Login
smash_ (because of cached stuff)
Home ! Login is what i have 22:52
audreyt the database thinks [particle] is as much as login as me
[particle] lemme commit
audreyt so must be cache at some layer
sure
smash_ does it send you a "special" link 22:53
after or before register ?
audreyt smash_: I invited [particle]
Limbic_Region pugscode.org down?
[particle] yes , i got an email message, with a special signup link 22:54
22:54 justatheory joined
[particle] from [email@hidden.address] 22:54
smash_ did audreyt received an invite too ?
svnbot6 r14674 | particle++ | s/\t/ /g
[particle] yay
smash_ different links ? 22:55
[particle] now, whot shall i invite? mestre.smash?
Limbic_Region pardon the confusion but 1. Is pugscode.org being down known and 2. How do I tell svn to point to the new repo permanently?
smash_ blushes.
[particle] lr: svn sw <old> <new>
wait, i think i'm missing an option there... 22:56
TimToady --relocate?
[particle] that's the one
Limbic_Region doesn't see a relocate option in help
smash_ --relocate : relocate via URL-rewriting
$ svn help sw
Limbic_Region ahh 22:57
ok - and the old repo was at what url?
audreyt svn sw --relocate svn.openfoundry.org/pugs svn.pugscode.org/pugs
[particle] audrey, what next?
obra audreyt: Set-cookie: JIFTY_SID_80
lambdabot Title: Revision 343: /
audreyt should do. try it?
obra port 9999 + a cookie for port 80 == sad
audreyt obra: ah. what should I do to fix that?
Limbic_Region thanks all 22:58
smash_ obra: ?
obra audreyt: is it running on 9999 intentionally?
Limbic_Region now about pugscode.org being down - known issue or just me?
audreyt obra: yes
[particle] obra: that might be happening with bitcard on rt.perl.org/rt3/ as well, then
audreyt Limbic_Region: known, will deal later
Limbic_Region k
smash_ Limbic_Region: no page here too
obra [particle]: no. that's different i think 22:59
audreyt: what port is in our jifty config file?
audreyt 8888
just changed to 9999
BaseURL: svn.pugscode.org:9999
lambdabot Title: Welcome to CommitBit for PugsCode
audreyt Port: 9999
should work now
obra audreyt: send me a commit bit and I'll test ;)
audreyt [particle]: retry login, then invite a new administrator 23:00
smash_ no httpd running on pugscode.org:80
[particle] ok, will try
audreyt [particle]: [email@hidden.address] under name "lwall"
[particle] Ctrl-<F5> to clear cache
Transferring data from svn.pugscode.org... (takes a while) 23:01
obra audreyt: in devel mode?
audreyt obra: commit bit sent
[particle] not working for me
audreyt obra: no, devel=0
[particle] so, i login => welcome page
click 'home' 23:02
click 'pugs'
click 'Admin page'
=> login page
i'll try another browser
audreyt hm. maybe it cached svn.pugscode.org:9999/admin
lambdabot Title: Manage projects and repositories
obra sJIFTY_SID_9999
audreyt can you try going in to commitbit.pugscode.org directly?
smash_ if it keeps sending you to "login page" it's because you aren't logged in
audreyt obra: as in, it works for you? 23:03
obra I get the right cookie. but i'm waiting on greylisting
audreyt in the "people" page it should display whether they picked it up or not
audreyt hacks
[particle] grr 23:04
obra audreyt: move unconfirmed people to a separate section?
Limbic_Region isn't in a big rush to get his commit bit and admin status back but doesn't want to be forgotten about either
audreyt Limbic_Region: all in due time :) 23:06
[particle] yeah, i can't get in no matter what 23:08
audreyt frowns
smash_ error_log ?
obra cookie issues?
audreyt Limbic_Region: can you try login?
23:09 ferreira joined
smash_ clean cookies 23:09
audreyt errorlog is clean
[particle] i just started exploder7 for the first time, cleaned cookies/cache, went to commitbit.pugscode.org (redirected to svn.pugscode.org:9999)
smash_ is access_log grwoing ? # just to make sure you're checking the right log
obra logged in ok 23:10
audreyt: shall I invite lwall?
audreyt sure go ahead 23:11
[particle] hrmm pugs project page looks very different in ie vs ff
People is below License, rather than right of Pugs
obra audreyt: that's throwing an error. 23:12
audreyt [Wed Nov 08 00:11:54 2006] [error] [client 66.92.67.231] FastCGI: server "/data/svn/commitbit/bin/jifty" stderr: ERROR - There was a validation error for person, referer: svn.pugscode.org:9999/admin/project/Pugs/people
lambdabot Title: Login, tinyurl.com/ss5yz
Limbic_Region audreyt - trying now
audreyt obra: so. you are not users.admin 23:13
but projectmemeber.level=administrator
obra indeed.
audreyt obra: that is the correct permission state, no?
obra looks like a logic bug in commitbit
(haven't had a lot of experience with non-superuser metacommiters)
audreyt whew.
obra sorry. working
audreyt k
Limbic_Region audreyt - not working AFAICT 23:14
audreyt Limbic_Region: as in, can't login? 23:15
Limbic_Region I tried both the email addresses that I might have registered with originally and my pw that was working as of yesterday - it says I might have typed something incorrectly
obra Limbic_Region: no. new url in your inbox
audreyt Limbic_Region: sorry, the old pw, we can't get plaintext
and md5 digest can't be made to work (yet) 23:16
Limbic_Region goes to check his inbox
obra audreyt: I'm almost 100% positive that it's ProjectMember.pm sub create with the wrong user.
audreyt so we decided to reinvite
obra but I'm falling over for lack of food
Limbic_Region is in a foul mood so is sorry if he is less than helpful
audreyt obra: ok. I'll deal.
Limbic_Region: aww.
obra audreyt: thanks. sorry
audreyt what happened?
Limbic_Region I have no email in my inbox 23:17
audreyt - in a nutshell, geographic distance is making me incapable of properly taking care of someone I love and who, to a certain degree, depends on me
audreyt oy. I know that feeling (only too well).
Limbic_Region: I sent to the _2000 one in yahoo 23:18
zptao you've done well
Limbic_Region audreyt - I checked that one and my gmail to include spam folder - nada 23:19
can you instead try Joshua period Gatcomb at gmail ?
audreyt Limbic_Region: sure. done 23:21
smash_ invite me too || :P j/k
Limbic_Region works great 23:22
so I won't have to change anything in svn for a ci?
audreyt provided you relocated
then yes. can you try a commit?
Limbic_Region sure 23:23
audreyt smash_: sure. email addr?
23:24 justatheory joined
audreyt Limbic_Region: can you try invite [email@hidden.address] as "lwall" under "administrator" priv 23:24
smash_ audreyt: [email@hidden.address] (thks)
audreyt in the commitbit.pugscode.org interface?
Limbic_Region audreyt - will do in just a sec 23:25
svn ci wasn't as friendly as I had hoped
audreyt smash_: done. 23:26
smash_: add yourself to AUTHORS and commit?
Limbic_Region I had to --username first timme
trying larry's commit bit now
seemed to work ok 23:27
svnbot6 r14675 | Limbic_Region++ | Added a link to perlmonks discussion on fib memoization example
Limbic_Region all things back to normall
audreyt good
whew
ok, I'll resend all the other commit bits now.
thanks to all for testing :)
smash_ audreyt: seems to work ok 23:28
audreyt: err.. i don't have nothing productive to ci :/
audreyt smash_: add yourself to AUTHORS 23:29
then ci that
welcome aboard btw :)
smash_ err, thks
let me co first
TimToady Looks good. not going to relocate till I ensure that openfoundry is synced up. 23:30
[particle] ok, things working better now, but js errors on the admin/project/Pugs/people page
jrockway did the repository uuid (etc.) change? if so I think you have to re-checkout 23:31
audreyt jrockway: no, we preserved revnnum and uuid
and all revprops
with svk/util/pullyu you can do that to any repo :)
basically cloning a remote repo
incrementally
jrockway cool! 23:32
23:32 Aankhen`` joined
jrockway i remember reading on the svk wiki that you had to re-checkout 23:32
obra just wait until you get pushmi. Then we can play games with which repo is master and you'll never need to care
audreyt yeah, pushmi and pullyu are New Technology
as of this week
smash_ couln't you have just rsync'ed it ' 23:33
?
audreyt and then the "k" part of "svk" will finally mean something. 23:34
smash_: rsync over mounted dav?
that is a curious thought, but I don't think it works that way
(no, we don't have shell access on openfoundry)
smash_ stop everything && umount everything && run live cd && rsync to new location && mount && start everything again
that could work ;) 23:35
(or not)
23:35 LimbicRegion joined 23:36 LimbicRegion is now known as Limbic_Region
jrockway tasty. now forking open source projects will be easier than ever ;) 23:36
no more asking people to recheckout their working copies, they can just svk relocate and "stick it to the man" ;)
Limbic_Region sorry, had access issues
audreyt yeah. just like All Other Modern VCS :) 23:37
Limbic_Region audreyt - did you want me to help redistribute commit bits?
audreyt Limbic_Region: no, I'm writing a script to do that
but thanks
smash_ (since i'm here, lemme build pugs ;) ) 23:38
audreyt reinvite seems to work
reinviting all admins... 23:39
all admins reinvited. reviting authors... 23:42
obra audreyt: what was the bug?
audreyt obra: current_user_can
but sorry, I'm on the brink of ENOFOOD fading
so let me finish reinviting first
23:42 bsb joined
obra okie. I'll look for a checkin post-food 23:43
audreyt I actually hardcoded it so any project admin become superuser.
obra oh
audreyt I mean, for the purpose of creation
you wouldn't want that.
but we only have 1 project
so *shrug*
obra nod
"first time I need a metacommitter" 23:44
23:44 cmarcelo joined
ingy I got mail from audreyt 23:44
audreyt good. :)
all authores being spammed... 23:45
cmarcelo ola. this new commitbit password is non-related to feather passwd, right?
obra correct
audreyt I really need to write up a faq and post to blogs and pugs homepage and p6a 23:46
Limbic_Region anyone care to speculate if chromatic is taking a jab at pugs in general or its many incomplete backends at use.perl.org/~acme/journal/31531
obra It's not worth speculating 23:47
Limbic_Region you're probably right
Limbic_Region wanders off to cool off 23:48
obra There are many ways to a running Perl 6. What's cool is that there is more than one way to do it.
audreyt and I replied to apologize. 23:51
23:52 ferreira left
Aankhen`` Oooh, interesting... ECMAScript 4... 23:54
Oh, 2008. Not so interesting then.
audreyt "production" 2008
they have part of it already rolled out in JS1.5 23:55
shipped with firefox2, I think
Aankhen`` FF2 had JS 1.7.
I guess full ECMAScript 4th Edition compliance is targeted at Mozilla 2.
clkao audreyt: you are evil
audreyt correct.
clkao: hm?
Aankhen``: indeed, I stand corrected :) 23:56
clkao the reply. don't be evil, okay? :p
smash_ (this Linking ... completely trashes my laptop)
Aankhen`` audreyt: Nah, you were right, you just got the minor number wrong. 23:57
Aankhen`` wonders why Mozilla still uses CVS.
audreyt svn switch --relocate svn.openfoundry.org/ svn.pugscode.org/pugs/ | run.pugscode.org | spec.pugscode.org | paste: sial.org/pbot/perl6 | pugs.blogs.com 23:58
Aankhen`` Habit? 23:58
audreyt probably.
eviltwin_b commitbit received 23:59
Aankhen`` I guess it might be difficult to convert the entire repository into an SVN repository.

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!