IRC channel logs
2016-07-23.log
back to list of logs
<amz3`>wingo: can I have multiple event loop in several thread with guile fiber? <jeffers-media>wingo: happen to take a look at my post a day ago? i was wondering why compiled elisp files are not supported <jeffers-media>fixing this could really help startup time, which is quite slow atm <galex-713>amz3`: oh, talking of guile-fiber, I searched on my search engines and didn’t find it :o <galex-713>PuercoPop: ok, but I don’t think I would be able to program a db with the perfs of sql dbs… :/ <galex-713>it seems these time increasingly more people are interested in guile-emacs… <janneke>ACTION also needs syntax-case, apparently even harder than syntax-rules, oh well <zv>any ideas of where to start if compiling breaks my code but `--no-auto-compile` succeeds? <ijp>it's not easy to give advice for that situation <ijp>if it doesn't depend on a lot of external stuff I can take a look <zv>when I do this, about half of my tests fail. <zv>When I extract thoses lines into a file called 'base' and `(load "base.scm")', it works with `--no-auto-compile` but not with regular guile <zv>well, I'm trying to make it smaller you see <zv>also, I highly value any general advice, be it broadly related to programming, lisp or guile specific. <zv>particularly from you ijp <ijp>load seems problematic <ijp>if you are mixing phases then breakage will happen <zv>i read the earlier document you linked about the tower, i thought that might have something to do with it <ijp>what if you make base a module <zv>ijp, the same number of tests fail as if I had just compiled my code <zv>let me put together a few branches to make life as easy as possible for you <zv>should you decide to look at it <ijp>with respect, I was a lot more interested when I thought this was an optimisation issue <ijp>I don't see any runnable tests in that repo <zv>ijp: it works fine when not loading <zv>ah, i forgot to push tests <zv>so, now when you run `guile --no-auto-compile sicp4.scm`, everything works fine <zv>when you run `guile sicp4.scm`, things break <ijp>what kind of errors are you getting <zv>a little tough to drill down because of the nature of the evaluator <zv>the point is that the code is exactly identical, but loading it causes an error <ijp>and I said `load' is a problem <zv>modules of course do the same thing <zv>depending on where I insert the `use-modules` in the file (of course always inserting it prior to any of it's exports being used) <zv>i.e, I get different numbers of test failures <ijp>your code is broken anyway <ijp>no eval-arithmetic.scm <zv>sorry, let me add that in, I just put that in a sec ago <ijp>okay, and I'm getting no bugs with autocompilation <zv>on master or external-evaluator-base? <ijp>nothing in the log files <zv>external-evaluator-base is where I use `load` <zv>`evaluator-base-as-module` is where I put it in a module <zv>i gotta use my grave symbols more judiciously <zv>in any case ijp, I appreciate you spending even 5 minutes on this, i'm sure this is painful to see for someone with experience with actually programming lisp <ijp>a lot of warnings, obviously, but no errors that I can see, and nothing in the .log files <zv>as far as I can tell, the srfi-64 default test runner doesn't always output failures to a log file <zv>you have to look in the stdout <zv>i've noticed that as well -- something will fail and the test files dont popular <zv>could that be a problem? <ijp>(except the warnings, and the obvious interpreter IO stuff) <zv>what version are you on? <ijp>stable before current <zv>im building stable 2.0 now <zv>everything works great now <zv>sorry to kerfuffle your time <ijp>yeah, be careful when running non-stable guiles <ijp>maybe there is an issue with 2.1, maybe not, but we'll need to reduce the example much further to figure that out <zv>so, in regards to modules <zv>any way to redefine a module's internal functions later? <zv>i.e if you have (module A (def fn1 (fn2)) (def (fn2) 2)) & you (use-module (A)) and later (def (fn2) 3) and then call fn1 <zv>is there any way to have fn2 override the definition inside of module A? <ijp>you can change those bindings, yes <zv>but it seems to me that fn1 will call the fn2 inside it's own module <zv>(at least from my bugs) <zv>even if I define fn2 later <ijp>yes, you need to define it in the right place <ijp>at the repl, I'd just switch to the original module with ",m" and redefine it the <ijp>in actual modules, you generally shouldn't be doing this, but you can with module-set! <zv>is there no way to manipulate the internals of a module? <zv>without EXPLICITLY doing so <zv>i.e, late binding of functions? <ijp>if we don't count goops, and I wouldn't, then not really <ijp>ACTION tests something <janneke>ACTION has given-up on syntax-case for now, and PEG, now trying to get records and lalr up <adfeno>I have a noob question, how do I import a library? E.g.: Mailutils? <paroneayea>I had a minor surgical procedure and was afk for a couple days