Parrot 3.4.0 released | parrot.org | Log: irclog.perlgeek.de/parrot/today
Set by moderator on 17 May 2011.
soh_cah_toa and honestly, unless you're playing a video game while editing a movie while you're running photoshop while you're using firefox (new firefox), you never will reach that 4GB limit 00:01
sorear whiteknight: the assertion that Ubuntu is "one of the most usable" is oft made but rarely substantiated 00:07
cotto_work seen dukeleto 00:09
aloha dukeleto was last seen in #parrot 1 days ago joining the channel.
00:25 theory left 00:27 darbelo left
plobsing bubaflub: pong 00:31
bubaflub plobsing: could you briefly describe the difference between 'p' and 'P' NCI types?
plobsing 'P' passes the PMC* through unaltered. 'p' uses the {get,set}_pointer convention for representing pointers with PMCs. 00:32
bubaflub plobsing: ok. i think i get it. 00:35
00:35 cotto joined
cotto ~~ 00:42
bubaflub plobsing: so in Winxed, if i have some var ptr set to an allocated StructView, when i pass that ptr into an NCI function should that be a 'p' or a 'P'?
if that question even makes sense 00:43
sorear do you want the NCI function to get the struct, or the PMC*?
should the passed pointer point to a Parrot_Obj header, or to your own struct? 00:44
plobsing ^^
sorear plobsing: when does 'p' use set_pointer? 00:45
plobsing sorear: for returns
and pass-by-reference
which are treated like returns by the NCI layer
sorear How do returns know what kind of PMC to make? 00:46
plobsing they don't. they just make Ptr PMCs
I suppose one might conceivably want to HLL map that 00:47
sorear does NCI ever set_pointer a non-Ptr?
plobsing no. What situation would make that desirable? 00:48
sorear none; I just was getting clarification
bubaflub plobsing: hmmm, i'm still a bit foggy on all this. almost all my functions take an mpz_t which i have for the StructView. i've got it working with a 'p' NCI type but i'm not totally sure that's right. 00:56
plobsing unless your function takes a 'PMC*' (or in rare cases a passthrough 'void*'), you want 'p'. 00:58
bubaflub plobsing: ok, that makes sense. i saw some NCI signatures in the extra_thunks file that took a 'P' and was wondering why that was. 01:00
plobsing: so those functions have to be designed to *explicitly* take a PMC*
plobsing yes, exactly
such functionality is generally only useful for calling into parrot internals or parrot extensions 01:01
bubaflub plobsing: great. also, pdd16 is still in draft mode - is there anything that's stopping that from becoming a non-draft PDD? more docs or updates?
(i'm willing to put in the time to edit it or fill it out, just curious)
01:01 theory joined
plobsing it looks like PDD16 describes parrot's NCI circa 1-4 years ago. 01:03
bubaflub plobsing: bummer. 01:04
plobsing: well, i'm still willing to edit it to get it up to date but i'll need some pointers in the right direction 01:06
sorear isn't that the way PDDs always are?
plobsing bubaflub: I'm not exactly the best person to ask about writting PDDs. my view is that they are ill-conceived and are doomed to uselessness. 01:07
bubaflub sorear: yes, but i'll put some time to fix at least this one
plobsing I think we'd be much better off with docs that didn't try to define "what should be", but in stead, focused on "what is"
and documented that in a sequence meaningful to users, not devs 01:08
bubaflub plobsing: for the most part i agree - i'd love to see the PDD at least get to the point where it's describing what currently is
plobsing but that's not a PDD. a PDD is a waterfall-design-type thing. 01:10
I'd encourage you to look at writting up a "How to work with native code from Parrot" document in stead, if you have the time. 01:11
bubaflub plobsing: yeah, i'd like to incorporate that as part of my GSoC project - a tutorial documenting the steps i took to get these bindings to work 01:12
01:14 pjcj left
bubaflub plobsing: but i think my project is pretty simple and has some idiosyncrasies that won't be too relevant to others 01:14
plobsing you can use your discretion about inclusions and omissions, but perhaps those idiosyncrasies are simply "advanced material" 01:15
which would go at the bottom 01:16
bubaflub plobsing: yeah. i have some specific scripts which "parse" (fsvo parse) the header and the docs to generate the code that i use. that's specific but the techniques are good. generating an extra thunk is something i think others will need. 01:19
01:44 whiteknight left 01:53 kid51 joined
kid51 msg dukeleto Something is amiss with the way planet.parrotcode.org is handling your latest post. 01:54
aloha OK. I'll deliver the message.
02:00 davidfetter left 02:05 bubaflub left 02:07 lichtkind_ joined 02:09 lichtkind left, lichtkind_ is now known as lichtkind 02:16 lichtkind left 02:20 benabik joined 02:22 kid51 left 02:24 woosley joined 02:29 rurban_ joined 02:32 rurban left, rurban_ is now known as rurban
benabik Good evening! 02:33
soh_cah_toa tcurtis: ping
benabik Random Q: Where does the value of a lexical get stored? In the register in the .lex declaration? 02:34
soh_cah_toa i think its the LexPad pmc 02:35
or LexInfo, one of those two
benabik: look at docs/book/pir/ch06_subroutines.pod 02:37
particularly in the "Lexicals" section 02:38
benabik The LexInfo doc says that a lexical is an alias for a register.
soh_cah_toa look at that pod file. i think that section is exactly what you're looking for 02:39
benabik soh_cah_toa: Well, I'm looking at the HTML on parrot.org.
soh_cah_toa benabik: LexInfo stores compile time info about lexicals while LexPad stores runtime info about lexicals 02:41
i'm not sure what you're trying to do but i don't think you can access those two pmc's directly from pir code 02:42
02:43 pjcj joined
benabik soh_cah_toa: I'm pondering possible optimization. If the register in the .lex declaration is the actual storage location, then there's no need to use find_lex in the same sub it's declared in. If it's stored in the LexPad... well, then, I'm sad. 02:45
soh_cah_toa: Also just trying to understand register use and whatnot because I'm playing with PAST's register allocation.
(By playing with, I may mean "screwing up"... Getting weird compilation issues.) 02:46
soh_cah_toa alright
benabik: www.onyxneon.com/books/parrotdev/pa..._guide.pdf 02:52
benabik: page 41 02:53
it's outdated but i still read it sometimes :)
benabik soh_cah_toa: This seems identical to docs/book
soh_cah_toa it is 02:54
dalek rrot: 3874860 | petdance++ | src/pmc.c:
consted a pointer
 03:14
benabik I have the sudden urge to make a crickets IRCbot, that just sits in the channel and chirps after a given amount of inactivity. 03:15
soh_cah_toa yeah, today has been unusually quite. thursday normally sees a lot of activity 03:52
github.com/parrot/parrot/graphs/punch_card 03:53
04:13 benabik left 04:27 theory left 04:41 benabik joined
dalek rrot/soh-cah-toa/hbdb: 0f058bc | soh_cah_toa++ | src/hbdb.c:
Renamed a few structs by adding the prefix 'hbdb' to avoid possible name collisions in the future
 04:46
rrot/soh-cah-toa/hbdb: bb3a61c | soh_cah_toa++ | src/hbdb.c:
Re-organized a few comments in src/hbdb.c to be more consistent with Parrot conventions
rrot/soh-cah-toa/hbdb: 13887fd | soh_cah_toa++ | src/hbdb.c:
Added skeleton for parse_command() and renamed hbdb_command_line() to command_line() because static functions shouldn't have the hbdb prefix. Also made a perldoc section for static functions
rrot/soh-cah-toa/hbdb: 42fc16a | soh_cah_toa++ | / (3 files):
Added src/embed/hbdb.c to the SEE ALSO section in all relevant perldoc
04:50 soh_cah_toa left 05:37 fperrad joined 05:39 athomason left 06:16 SHODAN joined 06:48 Drossel left, Kulag joined
dalek p: af3db5b | pmichaud++ | src/6model/reprs/P6opaque.c:
Improve NYFI 4 message to include name of attribute not located.
 06:57
07:26 PacoLinux left 07:27 PacoLinux joined, mj41 joined 07:39 woosley left
dukeleto ~~ 07:44
cotto nice to see you alive, dukeleto 07:45
dukeleto cotto: took a midweek vacation
cotto getting relaxed before OS Bridge?
dukeleto cotto: just in general 07:46
cotto: surprised to hear about perl-ctypes not working. I thought it was going into p5p core
cotto orly? That'd be amazing.
08:20 dodathome joined
dalek Heuristic branch merge: pushed 29 commits to parrot/leto/embed_grant by leto 08:28
08:49 Drossel joined 08:50 Kulag left 08:59 Drossel left 09:10 mikehh left
dalek website: rohit_nsit08++ | CorellaScript : AST and LALR parsing. 09:20
website: www.parrot.org/content/corellascrip...r-parsing.
09:21 Kulag joined 09:23 mikehh joined 09:29 SHODAN left 10:29 rurban_ joined 10:32 rurban left, rurban_ is now known as rurban 10:59 dodathome left 11:05 redicaps joined 11:06 SHODAN joined 11:08 lichtkind joined 11:33 mtk left 11:39 mtk joined 11:59 jsut_ joined 12:04 jsut left 12:23 contingencyplan left 12:26 cosimo joined 12:30 redicaps left 12:52 estrabd left, estrabd joined 12:54 rohit_nsit08 joined
rohit_nsit08 hello #parrot 12:54
12:59 bubaflub joined 13:05 darbelo joined 13:17 whiteknight joined
whiteknight good morning, #parrot 13:17
tadzik good morning whiteknight
whiteknight hello tadzik, how are you today? 13:19
tadzik not bad. I happened to fail my yesterday exam though
moritz :( 13:20
which one was that, math?
tadzik yes
whiteknight what subject in math? 13:21
tadzik I passed the exam itself, but I don't have enough points from the excercises: I wasn't present on the last of mini-exams, because I was writing physics then
it's named "Probabilistics and Statistics"
moritz that sucks, probably 13:22
tadzik if it wasn't for physics, I'll have my exercises passed too, and I'll be drinking right now. But in the situation I am, I need more points from the exam than the passing border
13:23 plobsing left
whiteknight tadzik: algebra-based or calculus-based probability? 13:23
tadzik this semester is a semester of second-attempt exams
whiteknight: I don't think my language interoperability applies to maths :) But I'll vote for algebra based, I think 13:24
there are matrixes :P
no, honestly, I don't know :) I can show you the tasks, but they're in polish 13:25
whiteknight tadzik: okay, that's the most common. I had it with a lot of calculus for the engineering curriculum. I don't recommend that for anybody
but probability is a tricky subject to wrap your mind around no matter what. Lots of formulas to memorize and all that 13:26
tadzik covariation. Do you know such a term? 13:27
oh, we have the card of formulas with us that's not a problem
calculus is a problem. I approach the woman today and ask "why do I have 3/10 points for this task?" – "well, a calculus mistake over here, so the eventual solution is wrong" – "so I don't get no points for the good method?" – "You get, 3 points" 13:28
out of 6 tasks, I couldn't really solve one, the rest was not a problem. And still I get 31/60 points. Calculus mistakes, typos, mistakes with integrals... Sigh. 13:29
PerlJam tadzik: make fewer mistakes :)
tadzik I will have to, yes :) 13:30
moritz tadzik: I know that all too well. I usually had all the methods correct and all the results wrong
PerlJam tadzik: do you usually take the whole time to work the problems?
tadzik PerlJam: very rarely, but this time I did 13:31
there were a few tasks I saw for the first time in my life, and I needed to figure out the way to solve them
otherwise I'll re-check everything, yes 13:32
I usually do
bah, whatever. 29th is the day then
your favourite Signal Theory, whiteknight, is on monday
whiteknight tadzik: yeah, I loved that course. I took several like that. 13:33
signals, communications, filters, random signals, adaptive filters, etc 13:34
tadzik I don't really share the love, to say it gently. But >half of the exam is information theory, which I have no problems with 13:35
13:35 darbelo_ joined 13:37 darbelo left
whiteknight that's fine, it's not for everybody 13:40
The first thing I learned when I became a teaching assistant is that the other students didn't enjoy the subject as much as I did 13:41
PerlJam drest: I bet they enjoyed it when it was over more than you did ;) 13:42
er, s/drest/whiteknight/
rohit_nsit08 whiteknight: hello 13:46
Coke show of hands. who can see: 3.bp.blogspot.com/-O5-gNs4YwCk/TfsC.../varrj.png 13:49
(i'm getting a DNS error. rohit_nsit08 is not.)
atrodo shows an image for me
rohit_nsit08 i'm able to see
i made it in google docs ;-) 13:50
moritz fine here
bubaflub fine here
moritz Coke: should I mirror it somewhere you can see it?
Coke moritz: eh. I'll check it out when I get home.
danke.
rohit_nsit08: remind me of this the next time I complain about broken images. ;)
tadzik fine here 13:51
arnsholt rohit_nsit08: Small nitpick on your blog post: LR isn't Left-to-Right, but Left-to-right Rightmost derivation
rohit_nsit08 Coke: sure ;)
whiteknight good morning rohit_nsit08
rohit_nsit08 aarnsholt: thanks, I have mentioned that it produces rightmost derivation using leftmost reductions 13:52
whiteknight rohit_nsit08++ # Nice blog post
rohit_nsit08 whiteknight: good morning
whiteknight: thanks. I'm working on toPIR functions associated with each of the nodes to get the PIR convertion
whiteknight: found cloud9 IDE a good tool for javascript development. I am able to debug code 13:53
arnsholt rohit_nsit08: Ah, right. Just me failing reading again. Sorry =)
13:53 woosley joined
rohit_nsit08 arnsholt: thanks for reading. I am new to blogging so any feedback is valuable :-) please free to comment and correct if I made any mistake 13:54
13:55 woosley left
whiteknight rohit_nsit08: oh good, I can't wait to see toPIR working! That's a very important part of the project 13:56
13:57 SHODAN left
rohit_nsit08 whiteknight: yup , I am struggling on some topics related to PIR, just discussed few with Coke and hope to get the solutions asap. 13:58
whiteknight rohit_nsit08: okay. I can probably help answer questions too. Let me know what you need 13:59
rohit_nsit08 whiteknight: Coke suggesting me some opcodes for read ( related to namespace ). After getting clear idea of questions and my approach. I'll put them on developer list tomorrow 14:01
suggested* 14:03
whiteknight okay, awesome. Namespaces can be tricky 14:07
14:08 lucian joined 14:20 lucian left, lucian joined 14:26 ambs joined 14:39 bluescreen joined 14:56 lucian_ joined 15:01 lucian left 15:16 dmalcolm joined 15:20 rohit_nsit08 left
dalek p: 5e83168 | pmichaud++ | src/6model/reprs/P6opaque.c:
Improve NYFI 1 error message with attribute name.
 15:29
whiteknight NYFI? 15:30
PerlJam Not Yet Fully Implemented 15:31
or ... you can choose another word for the F if you like ;)
cotto_work ~~
whiteknight okay 15:33
15:43 alester joined 15:50 rohit_nsit08 joined 15:55 mj41 left
dukeleto tcurtis: pong 15:56
alester blog.wikimedia.org/2011/06/13/serve...donations/ 16:12
Do we need boxen?
whiteknight I've got a boxen under my bed that I couldn't find a place to host
dukeleto whiteknight: what kind of boxen? 16:13
whiteknight Dell Poweredge something or other 16:14
I can't run it out of my apartment, because it's too loud 16:15
16:22 theory joined
atrodo whiteknight> I have a dell 1U in my basement. Difference is it's running (not quiet either) and I built a shelf just for it 16:26
whiteknight atrodo: I don't have a basement. I have an apartment the size of a volkswagon 16:28
when my kid generates too much wind, we hear it. The server will not go unnoticed 16:29
atrodo whiteknight> Then yes. When I was in the apartment when I first bought it, it was in a closet and was borderline too loud
alester I'll host servers. I've already got one. 16:34
Right now it's just running the odd build and splint. And folding@home. And a proxy server.
17:00 hercynium joined
whiteknight alester: where do you live? shipping on these hogs is non-trivial 17:02
at least, not if you want to use it for more than modern art on arrival
alester Chicago area.
whiteknight I'll have to price out shipping for that. 17:03
17:04 particle1 left 17:05 particle joined
whiteknight actually, we're in the process of buying a house right now, but I promised my wife I would throw the server away if I couldn't find a home for it by the time we moved out of the apt 17:06
atrodo whiteknight> 1U? Specs? 17:08
17:09 bubaflub left
whiteknight I think it's 1U. I have to look at the specs again 17:09
it was a hand-me-down
dukeleto has 4U's of servers collecting dust in his closet 17:11
and a 20U rack collecting rust outside 17:19
whiteknight if dust were worth anything, you could sell the collection on ebay
dukeleto whiteknight: can you tell me how one would go about adding coverage to the uncovered code in Parrot_ext_try ? tapir2.ro.vutbr.cz/cover/latest-c_c...end-c.html 17:21
whiteknight: it that code even reachable?
whiteknight oh wow, that's a doozie 17:22
dukeleto whiteknight: i have to cover Parrot_ext_try to hit 95% coverage in that file
whiteknight it's reachable, but...damn
dukeleto whiteknight: yeah, not simple. Do I have to install an exception handler and not pop it off? 17:23
whiteknight: i see that i have to mess with the current context, but I don't fully understand the code. I stared at it for about an hour last night.
whiteknight yeah, this is going to take some thought 17:24
The "do_panic" part is probably not reachable 17:25
unless somebody craps on the curctx pretty hard
you may need to call a Sub, create an inner runloop (through a vtable override on Object, etc) and throw an unhandled exception from there 17:26
We added those checks because of the IMCC refactor work, so I know it's necessary
I just cant remember the specific steps that caused us to trigger that
that's a great example of code that needs tests, precisely because of how complicated it is 17:27
dukeleto whiteknight: joy 17:37
17:39 dodathome joined
dukeleto whiteknight: well now at least I have an idea of what needs to happen, thanks 17:39
whiteknight It's entirely possible one of those two zones is not reachable. I suspect they both are, but you might have to get pretty malicious to do it 17:40
I *know* we added that code for the IMCC refactor work, so a test involving IMCCompiler PMC might not be a bad place to start
IMCCompiler pushes a context on the stack, to hold space for temporary registers. If IMCC throws a parse exception, it might not clean that up. 17:41
]I suspect we've addressed those problems, but maybe I am misremembering
if you're willing to get real down-n-durty, you can manually construct a faulty case
Parrot_push_context, then throw an exception, etc 17:42
dukeleto whiteknight: i think i have triggered this thing from PL/Parrot
whiteknight: but it was from an IMCC misparse
whiteknight: this looks shiny: github.com/tolmasky/language 17:44
msg rohit_nsit08 you may want to take a look at github.com/tolmasky/language 17:45
aloha OK. I'll deliver the message.
whiteknight who would have thought that something so simple as "add exception try/catch to C" would involve such complicated code 17:52
I mean, C++ does it, and that isn't complicated at all 17:53
:)
atrodo whiteknight> Where's your closing sarcasm tag?
whiteknight atrodo: burned by the heat of my burning wit 17:54
17:54 cosimo left
atrodo whiteknight++ 17:54
18:12 bubaflub joined
lucian_ whiteknight: i may have said this before, but cyclone's exceptions are rather simple 18:16
dukeleto bubaflub: wazzup 18:17
bubaflub hey dukeleto - not much, how are you?
dukeleto bubaflub: goodly 18:18
bubaflub dukeleto: great. my tests are coming along quite nicely - i've almost finished all the division tests
dukeleto: there three major kinds of divisions, round ceiling, round floor, and truncate 18:19
dukeleto: and then all the permutations for quotient, remainder, both; use an GMP integer or unsigned int, so on and so forth
18:21 rohit_nsit08 left
dukeleto bubaflub: mmmm, tests 18:22
whiteknight bubaflub: that almost sounds like something you could put into table form 18:23
and then have a loop that runs over each row, divide A by B using method C, expect result D 18:24
actually, I had wanted to add a similar mechanism to Rosella.Test to do parameterized tests like that 18:25
18:29 rurban_ joined 18:32 rurban left, rurban_ is now known as rurban
bubaflub whiteknight: i've seen that done with Cucumber 18:33
that is, parameterized testing
it can be very very nice
and yes, there is an awful lot of the code that could be refactored and simplified
create an Integer, set it to this, pass it to this function, get the result, check the result...
whiteknight I'll play around with that tonight 18:34
I've got a good idea for it in my head. Problem is making it work well and be pluggable 18:35
bubaflub whiteknight: i'm also thinking of maybe adding a script to my project to generate winxed test files like that 18:36
whiteknight ok 18:37
It will probably be something like Rosella.Test.test_data(<array>, <function>, <optional named args>) 18:38
will execute <function> as a test on each item in <array>
or, I guess <array> could also be a <hash> for "test name" => "test data"
that works fine for me 18:39
18:43 lucian joined
whiteknight bubaflub: How does something like this look for you? gist.github.com/1032022 18:46
18:47 lucian_ left
bubaflub whiteknight: ooooh, shiny 18:47
whiteknight: that's kinda what i was thinking, except way better 18:48
whiteknight: i also have a random Winxed question - what exactly does the "using" statement do? and when do i need / not need it?
whiteknight I think we're going to run into a hell of a lot of problems getting the function to be treated as a method, when it isn't declared as a method
bubaflub: that's a great question. I don't know everything it does 18:49
basically, it does a namespace lookup, and stores a value in a named variable
so "using Rosella.Test.test_vector" would look up the "test_vector" function in namespace ["Rosella"; "Test"], and store it in a local variable named "test_vector"
the "using static" sequence used to create a static reference to a named function in the same file. I don't think it does that anymore 18:50
"using extern" will do a library load, though I've never played with that so I don't know the details
bubaflub whiteknight: ok, that makes sense. 18:51
whiteknight and I think "using namespace" makes the contents of another namespace visible to the compiler, so you can use functions from that namespace in your code without a million little "using" statement for each function you want
but I think that only works at compile time
bubaflub whiteknight: ok, that makes a lot of sense.
whiteknight NotFound tends to put lots of namespaces and classes in a single file, so he uses those kinds of tools himself. I tend to write each class and namespace in a separate file, so I lose out on some fancy compile-time magic 18:52
18:58 mj41 joined
bubaflub whiteknight: yeah, my NCI bindings are in a PIR file under GMP::Raw 18:59
the winxed class GMP::Integer uses those, and then has functions that wrap those with sugar
like automatic conversion of C-strings <-> Parrot strings, type checking, constructor and destructor 19:00
and eventually some vtable overrides
whiteknight I think there is a syntax for putting in forward-declarations, so you can use "using namespace", but I haven't played with that at all either 19:01
19:07 bluescreen left, bluescreen joined
bubaflub whiteknight: ok. since my Winxed code is generated, i've just got a bunch of 'using ...' statements 19:09
19:11 mj41 left 19:13 mj41 joined 19:14 hercynium left 19:37 contingencyplan joined 19:38 dmalcolm_ joined 19:42 dmalcolm__ joined 19:43 dmalcolm left 19:46 dmalcolm_ left 20:10 mj41 left 20:14 jevin left 20:27 perlite_ joined 20:30 whiteknight left 20:31 perlite left, perlite_ is now known as perlite 20:49 dodathome left 21:33 fperrad left 21:42 PurityControl joined 21:46 Patterner left, PurityControl is now known as Patterner 21:47 ambs left
dalek rrot-gmp: 29123cb | bubaflub++ | t/integer/div/ (3 files):
add coverage for the 27 division functions mpz_cdiv*, mpz_fdiv*, mpz_tdiv*:

mpz_cdiv_q, mpz_cdiv_r, mpz_cdiv_qr, mpz_cdiv_q_ui, mpz_cdiv_r_ui, mpz_cdiv_qr_ui, mpz_cdiv_ui, mpz_cdiv_q_2exp, mpz_cdiv_r_2exp, mpz_fdiv_q, mpz_fdiv_r, mpz_fdiv_qr, mpz_fdiv_q_ui, mpz_fdiv_r_ui, mpz_fdiv_qr_ui, mpz_fdiv_ui, mpz_fdiv_q_2exp, mpz_fdiv_r_2exp, mpz_tdiv_q, mpz_tdiv_r, mpz_tdiv_qr, mpz_tdiv_q_ui, mpz_tdiv_r_ui, mpz_tdiv_qr_ui, mpz_tdiv_ui, mpz_tdiv_q_2exp, mpz_tdiv_r_2exp		 22:03
22:11 bluescreen left 22:15 soh_cah_toa joined
soh_cah_toa ~~ 22:15
bubaflub evening soh_cah_toa 22:16
soh_cah_toa bubaflub: hey
bubaflub how goes the debugger? 22:17
dukeleto bubaflub: nice tests
bubaflub dukeleto: thanks
soh_cah_toa bubaflub: i'm on the verge of a breakthrough. i'm very close to command recognition. working on parsing input 22:18
bubaflub soh_cah_toa: that's excellent.
soh_cah_toa yup
then i have to look at some test examples for ideas. i need to test that feature but i can't figure how to fake input since it's an interactive program 22:19
bubaflub soh_cah_toa: one way to do it is if you have a separate function that parses the input, test only that
soh_cah_toa bubaflub: yeah, i thought of that but it's a static function 22:20
bubaflub soh_cah_toa: let's say you've got a function like (psuedocode): string parse_input(string input);
soh_cah_toa: is there source i can look at? i may be able to help 22:21
soh_cah_toa sure, let me look quick. i think i have to push
dalek rrot/soh-cah-toa/hbdb: 0b4c473 | soh_cah_toa++ | src/hbdb.c:
Beginnings of parse_command(). Not yet complete
 22:23
cotto_work soh_cah_toa: you should be able to fake input and output to the debugger by messing with stdin, stdout and stderr from the Perl testing code.
write a function to start the debugger with a given set of arguments and another to run a single command and check that it returns the right output 22:24
soh_cah_toa bubaflub: github.com/parrot/parrot/blob/0b4c...src/hbdb.c it's the parse_command() function around line 410 (it's not finished yet) 22:25
soh_cah_toa thinks about this
cotto_work testing the output of a static function won't be fun
soh_cah_toa wonderful
bubaflub soh_cah_toa: here's what i'd do - i'd break out the parsing of the input from the commands into a separate function 22:27
cotto_work I highly recommend an alternate approach.
soh_cah_toa the thing that i'm struggling to figure out is how do i actually "reach" that function? how can i call a static function in a c source file from a completely unrelated perl script?
cotto_work: approach to what?
bubaflub soh_cah_toa: the perl script can compile a simple C program that #include's your file
cotto_work testing C functions directly as opposed to testing them by having the test suite running the debugger 22:28
bubaflub soh_cah_toa: it ain't pretty but it's possible
soh_cah_toa hmm...ok
so i could pass a command on the command-line. like w/ `` or system() 22:29
but my program doesn't do that
22:29 alester left
soh_cah_toa wouldn't it make sense to test the program exactly how it would be used by the user rather than defining some special condition in my program for testing? b/c then it's testing a different behavior 22:30
cotto_work soh_cah_toa: that's exactly what I'm recommending 22:31
soh_cah_toa alright
but then there's still the matter of automating a pretend interactive session 22:32
hmm...i wonder if i could do it w/ pipes like | 22:33
cotto_work You'd have to keep the handles around, but I think it'd be doable.
soh_cah_toa for instance: open FOO, "echo 'command' | hbdb" 22:34
yeah, i don't see why that wouldn't work
cotto_work make sure to write up a minimal proof of concept so you understand what you need to do 22:35
soh_cah_toa i'm not sure i follow :/ 22:36
cotto_work search.cpan.org/~jesse/perl-5.14.1/...C/Open3.pm looks like what you'd want 22:37
22:37 preflex left
soh_cah_toa it looks like that may help. however, then i'm adding another dependency 22:39
cotto_work dukeleto may be able to suggest a better alternative
22:41 preflex joined
dukeleto cotto_work: better alternative for what? 22:42
cotto_work dukeleto: is IPC::Open3 the best option for testing a debugger by writing to stdin and reading from stdout and stderr on the same process? 22:44
sorear I recomment IO::Pty::HalfDuplex personally 22:45
soh_cah_toa i think i'm headed in the right direction. however, when i execute "echo 'foobar' | ./hbdb hello.pbc", i get "ERROR: Null PMC access in get_string()" 22:46
sorear soh_cah_toa: readline probably returns NULLPMC for EOF
soh_cah_toa oh, i see 22:47
i guess that method won't work then. IPC::Open3 does seem like it's what i'm looking for but i'm still worried about dependencies 22:48
sorear no, you need to fix hbdb
user pressing ^D at the prompt shouldn't cause a crash!! 22:49
22:49 darbelo_ left
soh_cah_toa i didn't though 22:49
sorear echo '
echo 'foobar' is exactly the same as typing foobar, enter, control-D 22:50
soh_cah_toa oh, ok
cotto_work dukeleto: is IPC::Open3 in core?
sorear has been since 5.0
IO::Pty::HalfDuplex isn't in core yet though :( 22:51
soh_cah_toa well that's great. except now i need to figure out why ^D causes it to crash ;) 22:52
22:52 darbelo joined
sorear check your input loop 22:52
are you doing "isnull" after readline? 22:53
soh_cah_toa no
sorear oh 22:54
readline returns '' if EOF
but also sets eof
well, check the backtrace
22:56 dmalcolm__ left
soh_cah_toa so if i have "Parrot_pcc_invoke_method_from_c_args(interp, stdinput, readline, "S->S", prompt, &input)" i need to check "input" for null? 22:57
sorear only if you think the method might return null! 22:58
soh_cah_toa you said it did 22:59
sorear no 23:01
15:54 <@sorear> readline returns '' if EOF
soh_cah_toa aren't empty strings equal to null? 23:02
sorear not at all
soh_cah_toa wait, so then what does isnull have to do w/ it then? by which i thought you meant PMC_IS_NULL 23:04
dukeleto github.com/tanoku/spym : MIPS R2000 Virtual Machine written entirely in Python 23:11
23:23 mtk left
cotto_work dukeleto: interesting 23:25
TiMBuS now to port python to mips r2000 23:36
23:42 lucian_ joined 23:44 darbelo left, lucian left 23:49 whiteknight joined
whiteknight good evening, #parrot 23:53
cotto_work goes home
23:54 Kulag left
soh_cah_toa whiteknight: hey 23:54
23:55 Kulag joined
whiteknight hello soh_cah_toa 23:57
23:58 Kulag left, Drossel joined
soh_cah_toa whiteknight: your mouse still acting up? ;) 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!