***fangism is now known as fangism-hongru
***fangism-hongru is now known as fangism-hongry
<civodul>"Compute the maximum fixed point of the data-flow constraint problem." *wingo learned quite a bit while playing with that stuff <civodul>cky: cky-hygienic-macros seems like a good idea <wingo>hehe, there are many things i still don't get :) <cky>civodul: Glad you like! :-) <cky>civodul: Oh, BTW, I saw something lately (that has your name attached to git blame) that I'd query you about. <cky>In the current documentation, in the Block Comments section, it says that (+ #| this is a #| nested |# block comment |# 2) evaluates to 3. <cky>Either there's a missing 1 before the block comment, or the result should be 2. <cky>dsmith-work: Oh, I see. ;-) <cky>Like a 2.99999999999999999999999999999999999999 kind of heavy 2, I guess. ;-) <civodul>cky: then the documentation may be broken :-) <ijp>no, mathematics is broken <ijp>actually, I believe I mentioned this before, but I do recall using one scheme implementation that allowed you to define numbers <ijp>I want to say it was stklos, but if it was stklos it doesn't do it anymore <cky>mark_weaver: Will do. <cky>mark_weaver: Also did you see my message on the mailing list about the broken (help "foo")? I'll be happy to push my change if it looks reasonable. <dsmith-work>cky: As in (+ 2 2) -> 5 for certain large values of 2 <fangism>hi all, really old problem on x86_64-darwin11 (seg fault): env GUILE_INSTALL_LOCALE=1 GUILE_AUTO_COMPILE=0 ../meta/uninstalled-env guild snarf-check-and-output-texi < objprop.doc <fangism>anyone else get this on guile-2.0.9? [guile-procedures.texi] Error 1 <fangism>2.0.7 did not have this problem, not sure about 2.0.8 <wingo>fangism: i can't reproduce that here with stable-2.0 git <fangism>objprop.doc is the smallest .doc that triggers seg fault <fangism>what's a good way to run the uninstalled executable in gdb? <wingo>i think my objprop.doc is the same <fangism>apart from relative paths to .c file, yeah <wingo>fangism: :) you can also meta/uninstalled-env env GUILE=meta/gdb-uninstalled-guile guild ... <wingo>terribly verbose but it does work <wingo>meta/guild should just work, that would be nice <fangism>hmm, still struggling to get gdb to find the executable... <fangism>gdb doesn't like guild (is a script) <fangism>is there an equivalent way to load guild from within guile? <wingo>then apply that module's main to the args that you choose <fangism>I did this so far: env GUILE_INSTALL_LOCALE=1 GUILE_AUTO_COMPILE=0 ../meta/gdb-uninstalled-guile <fangism>which booted into gdb, run then gave me the standard guile prompt <fangism>or from gdb prompt, can I run with arguments like -s ../meta/guild ... ? <wingo>then (use-modules (script compile)) <wingo>then ((@@ (script compile) main) "arg1" "arg2" ...) <fangism>arg1="guild", arg2="snarf-check-and-output-texi"? (sorry if these are stupid questions) <wingo>i think it starts already with snarf-check-and-output-texi <wingo>it would be (scripts snarf-check-and-output-texi) in that case of course <wingo>you can see these files in module/scripts/snarf-check-and-output-texi.scm for example <fangism>after (scripts snarf-check-and-output-texi), how do I redirect in (<) an input file? <wingo>perhaps (with-input-from-file "objcode.doc" (lambda () ...call main ...)0 <wingo>otoh it's such a short file that you could copy-paste into the terminal of course <wingo>though maybe that's not the thing to do, given that it's maybe a locale issue... <fangism>it didn't like the first 'snarf_cookie' unbound variable <fangism>how can you tell it's a local issue? <wingo>i don't mean to say it's only your issue :) <wingo>just thinking it could be locale related <fangism>what should the procedure argument to with-input-from-file be? <fangism>snarf_cookie is not exported from snarf-check-and-output-texi, so i'm thinking of trying with-input-from-file <fangism>i've done: (use-modules (scripts snarf-check-and-output-texi)) (with-input-from-file "objprop.doc") (snarf-check-and-output-texi) -- then it hangs (waiting for input? not using CPU) <wingo>fangism: the argument to with-input-from-file should be a thunk <wingo>so snarf-check-and-output-texi without the parens <wingo>wow, that could be a good one <wingo>what C compiler did you use to compile guile? <wingo>the new clang or the old apple gcc fork? <fangism>looks like it finished "object-property" and then died somewhere in "scm_set_object_property_x" ? <wingo>looked to me like the tortoise and hare thing didn't work <wingo>can you mail bug-guile with that backtrace? <wingo>i suspect if you compile guile with a lower optimization level it will work <fangism>Apple LLVM version 4.2 (clang-425.0.28) (based on LLVM 3.2svn) <wingo>wow, i have guile building now with rtl... <wingo>it has done eval, boot-9, psyntax, vlist ... <wingo>but that's farther than it's ever gotten before <wingo>iterative contification is neat, i have it printing how many things it contifies at each round <wingo>usually takes 3-4 rounds to finish, but for peval it is taking longer <wingo>sometimes it drops down to 1 contification but then that contification opens up 3 more possibilities... <civodul>wingo: is it something that needs to be tweaked, to avoid infinite recursions and such? <wingo>but the question is, is it an n^2 thing or what <wingo>for peval it took like 30 rounds, over a couple minutes <wingo>i'm sure it could be optimized a lot <wingo>dominator computation and loop detection are being run there when it's not necessary <wingo>but still i think it's probably worth it <wingo>each contification is an important optimization <wingo>and peval is really important, so i'm willing to pay for it *fangism is curious what the tortoise and hare business is about <fangism>wingo: rebuilding with clang -g -O1, we'll see what happens <fangism>wingo: -O1 still crashes :( bootstrapping clang-release-3.3 now, will try that later *TaylanUB built Guile 2.0.9.x fine on OS X a couple months back. <fangism>TaylanUB: with what compiler/options? <TaylanUB>Nothing fancy IIRC. It's on my work machine and I'm at home so can't check right now.