Parrot 1.1.0 Released | parrot.org/ | 325 RTs left | Weekly Priority: Remove Deprecated Items
Set by moderator on 28 April 2009.
Infinoid Any errors before that? 00:00
I dunno msvc terribly well, but it's the first logical question to ask :)
00:09 AndyA joined
whiteknight nope, no errors that I saw before that 00:12
nopaste? 00:13
clunker3 pasta.test-smoke.org/ or paste.husk.org/ or nopaste.snit.ch:8001/ or rafb.net/paste or poundperl.pastebin.com/ or paste.scsys.co.uk/
purl nopaste is at nopaste.snit.ch/ (ask TonyC for new channels) or rafb.net/paste or poundperl.pastebin.com/ or paste.scsys.co.uk/ or App::Nopaste or tools/dev/nopaste.pl or at www.extpaste.com/ or paste.scsys.co.uk (for #catalyst, #dbix-class, #moose and others)
nopaste "Whiteknight" at 69.249.200.13 pasted "Configure failure on Win64" (17 lines) at nopaste.snit.ch/16451 00:14
whiteknight I wouldn't even be using msvc if I didn't have to, but it's the only free 64-bit compiler that I can find for windows
gcc and mingw don't support it yet that I am aware of 00:15
Infinoid oh, so this isn't during make, it's failing to configure 00:23
whiteknight right, configure-time 00:24
Infinoid well, there are reports in PLATFORMS for win64 with both mingw and msvc
so in theory, at some point this worked
whiteknight well, mingw is 32-bit only on Windows, so those reports are probably inaccurate
Infinoid hmm, maybe they compiled 32 bit binaries under a 64 bit OS, you mean? 00:25
whiteknight most msvc is 32-bit too, it took me a long time to find a 64-bit compiler
yes, that's what I think happened
that's what I used to be doing, unwittingly
Infinoid well, Configure.pl does have arguments like --verbose-step and --verbose=2
Coke msg allison - for me, 'make' fails trying to parse something that loads the dynops, so that's my test.
purl Message for allison stored.
whiteknight also explains why there are no Win64 smolder reports, because people on win64 systems are actually building 32-bit binaries
Infinoid --verbose=2 should show you what command line it tried to run 00:26
Coke msg allison - ah, misread. yes. if you build pir, you're using the opcodes, pretty much. the script "error" should explicitly call tcl_error, though, if you like.
purl Message for allison stored.
nopaste "Whiteknight" at 69.249.200.13 pasted "verbose configure output for Infinoid++" (30 lines) at nopaste.snit.ch/16452 00:27
whiteknight yeah, so that's a crazy output
Infinoid "__security_check_cookie 00:28
"?
purl rumour has it " is the quote char, not `
00:28 bacek_ joined
chromatic I thought he was assistant TO the project manager. 00:28
Coke (tcl builds) not exactly. 00:29
whiteknight Infinoid: Exactly, I've never heard of that function before 00:30
Infinoid whiteknight: from googling, it looks like it's used somewhere in the CRT or linker stub stuff that the compiler adds to your binaries.
which rev of MSVC is this? I'm not sure if I should tell you to upgrade or downgrade, but that seems to be the conclusion of most of the google hits. 00:31
whiteknight SDK 14, compiler 14.00.40310.41 for AMD64 00:32
I think this is the most recent version, but I cant downgrade because I need the 64bit compiler which I can't find in previous versions
Infinoid well, hmm 00:33
does a hello world fare any better?
whiteknight nope 00:34
Infinoid See if removing any of these from the linker line help: kernel32.lib ws2_32.lib msvcrt.lib oldnames.lib 00:35
Last time I used win32 was a long time ago, but I remember there being a lot of redundant fluff in the libraries we linked against.
s/win32/msvc/ 00:36
whiteknight okay, I think I got it
I linked in the library bufferoverflowU.lib, which is apparently necessary now
because microsoft is trying to protect me from my buffer overflow mistakes
Infinoid Ohnoes. 00:37
Those evil, evil coders.
whiteknight no we just need to see how badly the build goes
answer: badly 00:38
cotto darbelo, pong
Infinoid particle++ once told me that it fails horribly because win64 is LLP64
whiteknight LLP64?
Infinoid purl, LLP64 is www.unix.org/version2/whatsnew/lp64_wp.html 00:39
purl OK, Infinoid.
whiteknight okay, the build is failing because it can't find sal.h 00:40
darbelo cotto. Is there any 'best practices' document for dynpmc writing?
whiteknight i never heard of that one either
Infinoid I'm not sure if LLP64 was the exact one. But I remember differences in int sizes being the reason why linux/64 worked fine and win64 needing a lot of work.
cotto darbelo, they're mostly identical to normal pmcs.
Infinoid sal.h is msdn.microsoft.com/en-us/library/ms...S.80).aspx 00:41
shorten Infinoid's url is at xrl.us/beqwg6
darbelo Is there any 'best practices' document for PMC writing?
whiteknight darbelo: no
but it's probably something that should be written
cotto darbelo, no apart from existing code
s/no/not/
Infinoid Did we get rid of the unionval from PMCs? If so, docs/pmc.pod should be updated 00:42
cotto yup
darbelo Existing code documents 'existing practices' :)
whiteknight existing code documents 'previously existing practices' 00:43
cotto whiteknight, I'll finish updating that tonight.
whiteknight cotto: w00t 00:44
cotto I wish documentation were as fun as it is important.
whiteknight I actually have a lot of fun with documentation, I've certainly written a lot of it over the years 00:45
cotto Hmmm. Maybe it's just a mindset issue.
whiteknight I probably need to write more for parrot, however
probably 00:46
purl Really? Probably? Are you Certain it's not certain? Are you sure it's unsure? I think you need to look harder. Because I'm an asshole.
cotto being honest doesn't make it ok, purl--
purl cotto: what?
cotto stupid auto-karma-filtering bot 00:47
forget probably
purl cotto: I forgot probably
cotto probably
darbelo I have, right now, "ATTR decQuad number;". Is it sane to keep the whole union there or should that be "ATTR decQuad *number" and keep just a pointer? 00:48
cotto is that in svn?
(and the next question is obvious) 00:49
darbelo I think so.
cotto doesn't look like it
The current practice is to only store primitive types or pointers in ATTRs. 00:51
If what you have now works, it'd mean fewer mallocs.
darbelo But it also means that the accesor macros will do struct assignment, and to get a pointer you'll need to use Parrot_DecNum_attributes direcly. 00:54
which strikes me as a 'not a best practice'.
cotto If it were me I'd just use a pointer. 00:55
darbelo I'll add the * and commit then. 00:56
Be back in a few hours .
cotto k
darbelo commit went in with pointers. 00:58
01:02 Theory joined 01:03 hcles joined 01:06 bobke joined 01:26 kid51 joined 01:39 leto_ joined 02:10 silug joined 02:35 janus joined 02:40 MariachiElf joined
kid51 msg allison Does the revised patch submitted by gerd in trac.parrot.org/parrot/ticket/509 address the issues you raised in rejecting the first version of the patch? Thanks. 02:42
purl Message for allison stored.
02:47 eternaleye joined
Coke . 02:49
03:08 japhb joined 03:29 darbelo joined
darbelo Is there anything in parrot_config that tells me if I am on a big/little endian system? 03:41
03:42 tetragon joined
cotto darbelo, bigendian 03:47
you can look at what's in it in config_lib.pasm 03:48
ooc, why do you need to know?
darbelo Have to define DECLITEND for decNumber. 03:49
hcles hello bacek! 03:51
bacek_ hello hcles
Tene There's something pleasant about implementing scheme on PCT for the fourth time. 04:05
hello hcles! 04:06
hcles Hello all round!
darbelo Maybe it's all the mistakes you know you are not making.
Tene :) 04:07
cotto Tene, did you clobber your implementation again? 04:11
Tene cotto: No, but I figured it would be fun to do again. 04:12
To be familiar with all the steps for my presentation.
also, doing it with .ast instead of $()
cotto I like that change. It makes actions less opaque. 04:13
Tene nodnod 04:14
04:14 flh joined
darbelo cotto: What is the instantiate VTABLE for? All the pmcs I can recall looking at have a "dummy" one. 04:15
cotto darbelo, I think it's used to make an instance of a class. 04:17
I don't think it makes sense for most PMCs. 04:18
see src/pmc/class.pmc for a non-dummy implementation 04:20
darbelo Just checking, I'm starting to fill out the VTABLES on decnum based on what the GMP Big* PMCs do. 04:21
cotto (but you don't need to worry about it)
I highly recommend compiling early and often, even if you don't expect the code to be usable (or fully compile). 04:22
darbelo I'll keep it in mind. Now that the makefiles half-work it's easier to do that. 04:23
cotto yeah. it's a good motivation to keep usable makefiles around (which makes it easier for other people to jump in if needed) 04:24
Coke darbelo: do they not have one or do they just get default's?
04:24 Theory joined
darbelo Coke: who not have a what? 04:25
cotto darbelo, do you know how VTABLE function inheritance works? 04:27
nm. it's probably obvious
darbelo Coke: are you talking pmcs or makefiles? 04:29
Hmm. Ran into th first decNumber shortcoming. 04:40
cotto ? 04:41
darbelo Iwas looking at doing set_integer_native(), but decNumber only handles int32_t. 04:44
Parrot's integers can be wider on 64-bit arches. 04:45
cotto Yeah, but you can convert to a STRING in that case. It's not optimal, but it'll get the job done. 04:46
darbelo Sure. It's just "You can handle a million digits and you choke on a lousy int64_t ?" 04:52
Kinda disappointing.
cotto agreed 04:53
darbelo Oh, do the makefiles work for you? 04:54
cotto They don't build the decNumber files, but they seem to work for the PMC. 04:55
darbelo I'm working on a Makefile.in for src/ 04:56
cotto It might be cleaner to put decNumber's files in their own dir under src
darbelo Yes. 04:58
cotto It's amazing how easy it is to get distracted while trying to write documentation.
Infinoid Yeah. Shiny things just leap out at you.
cotto minimizes all windows in all other virtual desktops. 05:01
05:08 tuxdna joined
dukeleto cotto: good idea 05:28
purl dukeleto: Good Idea: Giving your dog a bath. Bad Idea: Having your dog dry-cleaned.
dukeleto purl, bad idea?
purl dukeleto: i don't know
darbelo purl likes animaniacs, apparently. 05:29
purl darbelo: what?
cotto It's like closing them, only they're still there. 05:36
Does anyone want to proofread some documentation for me?
darbelo I can do it. 05:37
nopaste "cotto" at 96.26.202.243 pasted "draft update of docs/pmc.pod" (131 lines) at nopaste.snit.ch/16453
cotto Just check that it makes sense is understandable. I did fact-checking as I wrote it, so all the information should be accurate. 05:38
feel free to nopaste an updated version if you find anything to fix
darbelo++
darbelo UnionVal is an obselete 05:40
cotto yup
darbelo should be obsolete
cotto fix'd
darbelo on "=item * The PMC needs metadata."C<need_ext> 05:42
what happens if you try to to add an attribute to a PMC declared without C<need_ext> 05:43
?
Jus asking, really. 05:44
cotto The PMC_EXT will be added to the PMC, its _metadata populated and the attribute added to the _metadata. 05:45
basically, Parrot dtrt
darbelo OK. 05:46
In "=head2 ATTRs and PMC_data()": 05:48
Shouldn't C<Parrot_Integer_attributes> be C<Parrot_x_attributes>? 05:49
Or is it meant as an example?
cotto I've fixed that to make it clearer how the name is generated. 05:50
nopaste "cotto" at 96.26.202.243 pasted "draft update v2 of docs/pmc.pod" (129 lines) at nopaste.snit.ch/16454
cotto There are also some pod changes. 05:51
nothing major
shower time. lmk if you find anything else. If not I'll commit in an hour or so. 06:02
06:12 uniejo joined
darbelo Looks good to me, but I'm not a native english speaker so I might have missed something. 06:15
cotto darbelo++ for input 06:21
darbelo cotto++ for documentation 06:27
cotto is not excited about the (at least) two additional pdds that'll need to be updated. 06:29
darbelo don't worry I'm sure you'll do a great job :) 06:34
06:35 amoc joined
darbelo New commit for decnum-dynpmcs. I *think* I'm done with the build infrastructure for now, let me know if it doesn't work for you. 06:36
moritz where does the repository live? 06:38
and why don't we have report bot in here? ;-)
darbelo google code. 06:39
purl google code is, like, code.google.com, run by Chris DiBona
moritz Infinoid: care t tell dalek to track code.google.com/p/decnum-dynpmcs/ ? 06:40
darbelo "svn checkout decnum-dynpmcs.googlecode.com/svn/trunk/ working-copy" will give you a read-only copy
cotto darbelo, trying now... 06:41
06:41 iblechbot joined
cotto darbelo, not quite. after make realclean (with the old makefile) and perl Configure.pl, the new Makefile doesn't think anything needs to be built 06:43
darbelo from the top-level?
cotto yup 06:44
dalek rrot: r38413 | cotto++ | trunk/docs/pmc.pod:
[docs] rewrite pmc.pod now that the UnionVal is history
cotto in src/decNumber and src/pmc it seems to work
darbelo "make realclean -> perl Configure.pl -> make " builds nothing for you? 06:47
cotto nope 06:50
make: `build' is up to date.
darbelo what sstem are you on? 06:51
s/sstem/system/ 06:52
cotto ubuntu Hardy i386
nopaste "darbelo" at 190.3.154.38 pasted "cotto: See if this helps" (31 lines) at nopaste.snit.ch/16455 07:02
cotto darbelo, that works 07:04
darbelo I hate make.
cotto How odd. "build" must be special.
darbelo build is a directory. 07:05
it checks to see if the 'build' directory has changed, instead of going to the 'build' target. 07:07
cotto of course 07:08
silly make
darbelo but I'm on xubuntu intrepid and it works either way. 07:09
I hate make.
07:09 amoc joined
cotto h8 07:10
darbelo Ah well, do we have a better name than 'compile' or 'the-target-formerly-known-as-build' ? 07:11
07:11 amoc joined
darbelo Since 'compile' is a bit of a misnomer there. 07:12
cotto 'pmc'? 07:14
also, the Makefile doesn't appear to check the timestamps on the right files.
make && make builds the same files twice
(which is a better failure mode than not building often enough, but still incorrect) 07:15
darbelo Make hates me back. It seems.
cotto better to deal with it now 07:17
moritz for gnu make you could say '.PHONY build'
but i don't think that's portable
07:18 amoc joined 07:56 riffraff joined
darbelo ok, if r10 doesn't solve then it won't get solved today. I've had enough make for this week. 08:25
moritz has the program already started officialy? 08:26
or is it still "community bonding" time?
darbelo Bonding. May 23 is the 'start' date. 08:28
cotto although starting during the bonding period is fine 08:29
moritz sure
cotto darbelo, looks like that solves it 08:30
darbelo Until I test it on BSD again :)
Oh crap. I don't even want to guess what could happen on windows with this. 08:32
cotto first make it work somewhere, then everywhere 08:35
we've got some guys on windows around here who could help 08:36
08:37 masak joined
darbelo Now that you mention it. What is everywhere this days? 08:39
Various Linux, *BSD, Windows, Solaris, OS X are the 'popular' stuff 08:40
is anyone building parrot on more obscure systems? 08:41
moritz isn't there a PLATFORMS file in parrot?
cotto we had someone build on aix near the end of '08
PLATFORMS is a good place to look 08:42
for that kind of information
darbelo There is some stuff missing there. 08:44
aix was built on '20080510' 08:45
cotto oh. I misremembered
darbelo There is no OpenBSD, for which there are smoke reports passing all tests. 08:49
Stuff like that. It look outdated. 08:51
08:51 amoc joined
cotto It depends on manual updates. 08:55
something automated would be smarter
darbelo Maybe based on smolder. 100% => Supported. 09:00
Enough rambling for today. 09:09
Sleep time.
cotto night 09:13
09:56 bacek joined
bacek good evening 10:08
Did I missed something crucial?
cotto nafaik 10:11
bacek good-good. 10:13
bah... pmc2c generates a lot of useless crap... I can't even use it as reference implementation :( 10:16
Who can explain why PMC's VTABLE methods declared PARROT_EXPORT?
cotto What happens if you remove it? 10:28
This is as good a time as any to see what we really need. 10:29
bacek cotto: nothing. Everything just works. 10:32
cotto orly
bacek Noone should invoke those functions directly. 10:33
cotto I'd mention it at #ps, but that's what I'd expect since it's all function pointers.
bacek Even inside parrot they called thru vtable pointer
And all generated *_ro_* functions are useless. Cause they can be replaced by *_ro_default_* 10:34
cotto: if you'll check pmc_undef.h there is a LOT of functions marked as PARROT_EXPORT. One problem - they doesn't exists... 10:35
cotto It seems like such a nice virtual machine until you dig into it a little. 10:37
bacek yeah... 10:38
Parrot is badly broken. OTOH it's fixable :)
cotto What's an example function? 10:39
a lot are useless stubs, but I'm not seeing any undefined ones
bacek cotto: (revoking sentence about non-existed functions) Looks like I messed with different header... 10:42
cotto I look forward to a Parrot where such things are improbably, although it may be fictitious. 10:44
s/bly/ble/ 10:54
11:20 kid51 joined
kid51 Smolder server down again? Could not upload report to Smolder at smolder.plusthree.com 11:20
HTTP CODE: 500 (Internal Server Error)
make: *** [smolder_coretest] Error 9
ping smolder.plusthree.com
PING smolder.plusthree.com (70.42.42.150): 56 data bytes
^C
--- smolder.plusthree.com ping statistics ---
65 packets transmitted, 0 packets received, 100% packet loss
purl 10 packets transmitted, 0 received, +7 errors, 100% packet loss, time 9008ms 11:21
dalek rrot: r38414 | jkeenan++ | trunk (7 files):
Applying patch supplied by bsdz in trac.parrot.org/parrot/ticket/577: add '--no-line-directives' switch to Configure.pl.
 11:22
11:26 tuxdna joined
bacek Another stupid question - how I can make readonly pmc? 11:41
Coke update the vtables to disable all the setters? 12:14
there is src/dynpmc/rotest.pmc, for whatever that's worth (it's not a great example)
Infinoid bacek: it's not a stupid question. there are some readonly variants for some pmcs, but that code has never made much sense to me 12:20
Coke there is no switch you can flip to say "now you're readonly", as I understand it. 12:21
12:21 particle joined 12:22 iblechbot joined
Infinoid There is a VTABLE_IS_CONST_FLAG bit, and VTABLE_HAS_CONST_TOO and VTABLE_HAS_READONLY_FLAG bits. They don't appear to mean the same thing 12:23
12:38 rg joined 13:10 ruoso joined 13:22 eternaleye_ joined 13:26 nopaste joined 13:38 gryphon joined 13:41 jhorwitz joined 13:44 Gerd joined 13:45 riffraff joined 13:54 particle1 joined 14:12 HG` joined 14:37 Andy joined 14:49 Debolaz joined 15:08 tuxdna joined 15:09 Theory joined 15:33 baest joined 15:42 contingencyplan joined
Coke bah. I can build from an installed parrot or not, but not both. 15:46
(the prefix of "inst" for some parrot_config vars is borken. 15:47
... and more besides. bother. 15:49
16:06 ruoso joined
Coke blows up parrot. 16:18
16:18 particle joined 16:21 particle1 joined
Coke *handle = Parrot_dlopen(path->strstart); ... tell me that's safe. :| 16:31
it's not safe to assume that a strings strstart contains a valid c string, is it? 16:34
rg doesn't think so
jonathan No.
It may have a null byte in there. 16:35
Coke src/dynext.c abuses this heavily. 16:36
(that is, more than once.)
jonathan :-(
Mucking with strstart is usually wrong.
Coke opened a ticket. 16:40
16:46 barney joined 16:47 flh joined
Coke anyone with darwin that isn't 10.4/intel? 16:56
17:01 flh joined 17:06 uniejo joined 17:45 uniejo joined
pmichaud Tene: ping 17:50
actually, I have to leave for a bit, so...
message Tene I have most of the things needed for .HLL in the master branch -- I just want to move the 'parrot'-specific stuff into their own set of files (to be included at the bottom of perl6.pir). I should have that shortly. Then we can start playing with .HLL in master. 17:51
purl Message for tene stored.
18:01 nopaste joined
dalek rtcl: r340 | wcoleda++ | trunk/CREDITS:
- add christoph
 18:08
Infinoid I'm thinking about how dalek should map source control usernames to karma points 18:14
I think it's going to end up being a huge lookup table, which I don't really want to manage myself 18:15
So, is it ok if it just parses some special page from the parrot wiki? That way people can update it themselves.
I'd make it poll the page once an hour to refresh its list, or something along those lines. 18:18
18:19 bkeeler joined
cotto Infinoid, maybe a special file in svn? 18:20
Infinoid That's fine too. Anything which can be directly linked should be easy to poll. Which svn? 18:21
In theory, the parrot svn is the only one which shouldn't need any aliasing. 18:22
particle Infinoid: can we update dalek's lookup table from here? 18:23
Infinoid hmm. maybe it should just search through CREDITS?
particle: We could do that
18:23 bkeeler joined
particle /msg dalek karma particle for pmichaud cotto infinoid ... 18:24
:)
Infinoid heh
cotto +1 for CREDITS, since it already exists and has a well-defined format
(and it'll prevent sneaky stuff like particle's idea) 18:25
particle well, it'll make the sneaky stuff public
Infinoid ok. We might need to add an "A.K.A." attribute type so the Coke listing knows he's also called "wcoleda"
if that doesn't already exist
particle don't spend too much time on fixing karma
it's better spent on fixing bytecode :) 18:26
cotto A: isn't used
particle in the linux credits format, A is unused?
Tene purl: msg pmichaud Great! With any luck, I'll have some time to work on it tonight, if I can get my PCT presentation done before that.
purl Message for pmichaud stored.
cotto in CREDITS it isn't
Infinoid karma is just one of the things in my list of dalek fixes, which is what I'm working on now 18:27
purl is just one of the things in my list of dalek fixes, which is what i'm working on now has neutral karma
Infinoid starting with the test suite.
so by asking this kind of stuff, I'm really just planning ahead. Thanks, guys 18:28
cotto (Mark Glines)++
heh heh
Infinoid Whatever I do won't be retroactive, unless someone gives me commit access to purl. 18:30
particle (test suite)++
is the bot in a public repo yet?
i'd love a commit bit 18:31
Util In the Parrot CREDITS file, (U) is "subversion username" 18:32
But you need a different field for "IRC nick"?
Infinoid the bot is just an unpack of the botnix tarball, and it's diakopter++'s beast
particle so A could be Aliases
Infinoid the rss parsers sort of overlay over that; I've been thinking of throwing them on github
when I do, I'll let you know and give you access. 18:33
Util (I) for IRC Nick would leave Aliases for future criminal^Wunderground contributors :) 18:34
What is karma used for here? 18:36
cotto used?
purl used is my best bet for this price. new is about $300+ each
18:37 davidfetter joined
Util Does the amount of anyone karma (that the bot keeps track of) affect anything, like (for example) who gets Ops, or who the bots obey? 18:39
s/anyone/anyone's/
karma Util 18:41
purl util has karma of 60
Util karma util
purl util has karma of 60
particle1 humans get ops 18:44
karma is only for vanity
purl is only for vanity has neutral karma
Infinoid Util: Aliases is exactly what we need, though. Dalek already knows that Coke is Coke, but it doesn't know that wcoleda is also Coke 18:46
Util particle1 & Infinoid: OK, thanks! 18:48
TimToady karma? 19:00
19:07 uniejo joined 19:38 darbelo joined 19:40 AndyA joined 19:52 masak joined
Coke wcoleda? 20:42
wcoleda is also Coke
cotto wcoleda? 20:53
purl hmmm... wcoleda is Coke
21:03 particle joined 21:11 bkeeler joined 21:18 Whiteknight joined
bacek good morning people of past :) 21:28
masak revels in the past
darbelo I'm not PAST, the parser errors out on me ;) 21:29
cotto bacek++ #pmc cleanup suggestions 21:48
bacek, the header generator acts strangely with src/pmc/integer.pmc 21:56
nopaste "cotto" at 96.26.202.243 pasted "minimal pmc that causes pmc_pct's header generator to choke" (3 lines) at nopaste.snit.ch/16460 22:00
bacek cotto: try add ? after <c_arguments> in <c_signature> 22:05
cotto: we need more tests... 22:06
22:08 particle joined
cotto bacek, that fixes it 22:09
I'll add a test and commit
bacek cotto: ok, thanks 22:10
cotto done 22:13
dalek rrot: r38415 | cotto++ | branches/pmc_pct/compilers/pmc (3 files):
[grammar] handle empty function args
bacek thinks that referencing generated header from system header is really bad idea...
PMC_cont in include/parrot/sub.h... 22:14
pmichaud bacek: we have to have PARROT_EXPORT on many of those functions or they don't compile properly under Windows 22:15
(TT #631)
at least, that's what I've been told.
bacek pmichaud: it's weird... 22:16
but we can try to remove signatures from header files. 22:17
pmichaud agreed, but apparently the linker on some windows compilers won't find the symbols unless they're exported.
bacek VC6?
purl somebody said VC6 was better than vc5
22:17 gryphon joined
pmichaud I don't know the exact details. 22:17
bacek ok. I'll try to find more information.
jonathan Please don't break Parrot on Win32... :-) 22:19
bacek jonathan: why not? :) 22:20
jonathan Because I don't want to have to revert anothr patch? ;-)
It's not just VC 6 though.
It's all Visual C++
There is no catch-all "just export everything" flag. 22:21
So everything that needs to be exported has gotta be labelled. :-(
bacek jonathan: can you try compile some pmc without signatures in header file? E.g. "clean" pmc_undef.h? 22:22
jonathan bacek: You mean just remove the lines in there? 22:23
bacek jonathan: yes
cotto If it ends up breaking the build, we should add a note to whatever generates that code. 22:25
jonathan bacek: I'm doing it for ResizablePMCArray rather than Undef. 22:26
Will have you an answer shortly... 22:27
I suspect it might not make the Parrot build itself fail (though it may), but IIRC it may be that any dynpmc that relies on inheriting from RPA might explode.
oh no
Parrot build breaks.
src\\pmc\\multisub.c(973) : error C2065: 'Parrot_ResizablePMCArray_clone' : undeclared identifier 22:28
bacek: Or did you mean just remove PARROT_EXPORT?
bacek: ah, that won't fly, because then we'd need to remove all the PARROT_EXPORT in the .c file too or we get an inconsistent linkage error. 22:30
cotto bacek, in body_part in actions.pm, what's the difference between $m and $m.name? 22:31
rg bacek: i think it would be preferable if you got your branch in a state where the windows people could test it
jonathan bacek: Anyways, I remember putting a bunch of that stuff in place, and can't imagine I woulda gone through the pain of doing so just for the fun of it. So I suspect there's a reason. :-)
However, it was 3-4 years ago, so don't ask me what the reason was. :-) 22:32
bacek cotto: $m is whole function. $m.name just name.
jonathan: fair enough :)
rg: it's far away... 22:33
cotto bacek, got it. name vs body
bacek Is there some Visual C++ available for free? 22:34
jonathan bacek: Yes. 22:35
www.microsoft.com/express/vc/
cotto bacek, www.microsoft.com/express/vc/
so close
jonathan That site has such a retro feel. 22:36
bacek jonathan, cotto, thanks :)
22:38 particle joined
cotto I wonder if they have a version for Ubuntu. 22:38
bacek gotta go. 22:40
purl EXCUSE ME, I HAVE TO GO WASH MY COMPUTER
bacek afk # @kids ~~ school, $self->dayjob
cotto bye 22:41
22:52 tetragon joined 23:04 Limbic_Region joined
cotto msg bacek Is there any reason the 'attr' sub in src/vtable_info.pir doesn't use optional params? 23:12
purl Message for bacek stored.
23:22 Ademan joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!