<nalaginrut>I'm reading rbtree again these days, and I realized that the deletion in rbtree.ss (dwells in guildhall) is wrong <nalaginrut>there's something vague (I can hardly say wrong) in the book <<introduction to algorithm>>, so it's painful while checking and testing the implementation <davexunit>grrr, goops generics don't work with record types <davexunit>this limitation really screws up one of my plans <davexunit>to use the generics feature without buying into the rest of the OOP stuff. <davexunit>wellp, guess I won't be using generics... bummer. <davexunit>I guess goops does special things to make <integer>, <string>, etc. classes <nalaginrut>I think it's different between record-type and class, although record-type has inheritance now <nalaginrut>davexunit: btw, it's different result of class-of between srfi-9 and rnrs record-type... <davexunit>I guess I'll just make a little record type for what I'm doing. <mark_weaver>sneek: later tell davexunit: it turns out that GOOPS generics will work with SRFI-9 records, but you can't simply use the first argument to 'define-record-type' as the class name. I don't know what's the best way to do it, but if you create an instance of the record type, then use (class-of <instance>) to get the class object, you can use that class object to define methods. <mark_weaver>sneek: later tell davexunit: it's awkward though. we should improve this. ***heroux_ is now known as heroux
***byte48 is now known as Guest14756
<davexunit>hello sneek, I hear you have a message for me. :) <sneek>Welcome back davexunit, you have 2 messages. <sneek>davexunit, mark_weaver says: it turns out that GOOPS generics will work with SRFI-9 records, but you can't simply use the first argument to 'define-record-type' as the class name. I don't know what's the best way to do it, but if you create an instance of the record type, then use (class-of <instance>) to get the class object, you can use that class object to define methods. <sneek>davexunit, mark_weaver says: it's awkward though. we should improve this. <nalaginrut>mark_weaver: I think there's no any possible to specify 'src' to the procedural Tree-IL while I'm using LALR parser, no? <nalaginrut>If the answer is negative, I'll use the sexp Tree-IL <davexunit>mark_weaver: thanks for that! I actually discovered that myself a couple of hours ago. I will see how easy it is to make define-method support record types. <zdavis>Hi all- I am trying to update the repl so that ,L without an argument gives a list of available languages to switch to. So far, I have a list just hard-coded. Does anybody have any suggestions for looking up the possible language choices reflectively? <nalaginrut>zdavis: I believe there's no such feature, since I've proposed a patch to list all the available languages, but it's never applied... <wingo>zdavis: dunno, find all locations that (language) could resolve as, then look in those dirs; see ice-9/boot-9.scm for details on module->path mapping <nalaginrut>zdavis: and you may use 'for-humans?' to detect the proper language ***dsmith-w` is now known as dsmith-work
<nalaginrut>sneek: later tell janneke I think your patch to lalr-scm added source-location info, which could be used in procedural Tree-IL 'src' slot. But I don't know how to use this new feature. Is there any document or tutorial? ;-) ***dje is now known as xdje
<pallagun>Are there any things I need to do get guile to allow me to (dynamic-link "mylib.so")? I built as described in manual section 5.16.4.3. I've placed the resulting .so in my LD_LIBRARY_PATH which I exported in my .bashrc. I swear this worked for me last week but now I keep getting a "ERROR: In procedure dynamic-link: file: "mylib.so", message: "file not found"" <pallagun>I can't seem to find what I've done to mess it up. <dsmith-work>shows you exactly what files and where it's looking.