<please_help>The ,bt backtrace is too shallow and too low-level, anything I can do about it? <please_help>for instance, only displaying trace from modules I whitelist, or simply displaying a longer backtrace <mark_weaver>please_help: ",help bt" will tell you about the backtrace command. note the #:width option <mark_weaver>I don't think we have anything to whitelist certain modules ***karswell` is now known as karswell
***michel_mno_afk is now known as michel_mno
*wingo almost done porting the optimizer over to cps2 <ijp>wingo: is there a specific order for arguments in cps? <wingo>ijp: you mean, order of evaluation? <wingo>because e.g. vector-ref takes two arguments, the first is the vector and the second is the index <wingo>of course order of operations is fixed already in cps <wingo>so, i guess not quite sure what you mean <ijp>$primcall and $call take a list of arguments (already evaluated) <ijp>are they ordered the same way as they would be in source <wingo>not sure if there are exceptions <ijp>I'll have another look, see if I messed up <wingo>is it a problem with a specific primitive? <wingo>but if you are wondering about implementation, see compile-bytecode.scm and vm-engine.c <wingo>there are all the cases there <ijp>okay, this was my own problem <paroneayea>used a javascript feature in some browsers so that the source locations in tracebacks and etc <paroneayea>were actually in the original language (in this case, wisp) <paroneayea>but apparently it's popular in some of the cross compilers <paroneayea>ijp: so, maybe useful to you to be aware of, or maybe you already were! <duud>Guile's texinfo manual contains multiple references to to the r6rs manual - like this one "For complete documentation, design rationales <duud>and further examples, we advise you to consult the “Standard Libraries” section of the Report (*note R6RS Standard Libraries: (r6rs)Standard Libraries.)." There is no corresponding texinfo file in Guile's repository but there is one for r5rs '/doc/r5rs/r5rs.texi'. Did r6rs.texi disappear somehow? ***michel_mno is now known as michel_mno_afk
<ijp>duud: no, it was never added afaik <ijp>and it would probably be four .texi files <duud>It would be nice if someone could add them. The person who wrote the manual obviously had them. <cmhobbs>i'm not sure that this is a guile question so much as a debian question but most any apt operation i do ends with this: /sbin/ldconfig.real: /usr/local/lib/libguile-2.0.so.22.7.2-gdb.scm is not an ELF file - it has the wrong magic bytes at the start. <cmhobbs>anyone know what that might be? guile seems to be working fine for me <cmhobbs>all my guile code still runs, i just didn't want something sinister to be happening under the hood :D <please_help>So I have only two points in my codebase that can set a tensor record's tensor-sync property to #t. I use print debugging to check whether it's doing the right thing and it is, without a doubt. Yet I somehow still end up with tensor types that can never be synched being synched. I thought it could be (guardian) garbage collection (which modifies device memory, which is what I use to get info about the tensor), but I <please_help>disabled it so that's not it. I realize not posting any code might makes it harder but is there anything I might want to look at? Also the backtrace isn't helpful, count argument seems to be ignored for some reason and I get an extremely shallow backtrace because of it. <dsmith-work>cmhobbs: ldconfig only expects .so's in the various lib dirs, and complains when it sees something else. <mark_weaver>cmhobbs: the problem is that the expectations of 'ldconfig' (everything in /usr/local/lib is an ELF library) conflicts with the expectations of 'gdb' regarding the locations of the *.scm file supporting gdb for libguile. ***karswell` is now known as karswell
<duud>ijp: yes thats exiting. I would like to learn from your source code. Is it available online? <ijp>duud: I will be pushing a branch Real Soon Now <ijp>but it won't be generally usable for a good while yet <davexunit>ijp: but will it be "production ready" and "web scale"? <ijp>oh sure, it's that already *paroneayea slacked over lunch and did chapter 2 of The Reasoned Schemer <paroneayea>and now, the little prover. So many books to read! *paroneayea tries running ijp's generated javascript in guile's js :) <ijp>yeah, but at least it runs in node <paroneayea>funny to read that right around the same time sourceforge is wrapping software in adware <unknown_lamer>but also heh, joeyh and I have wunderground in the same tile spot <ijp>tbf, you won't notice those ads after a week or two <davexunit>but they shouldn't be there in the first place. *unknown_lamer has first hand experience <unknown_lamer>davexunit: I see you have the new Ultimate Edition XChat(tm) <davexunit>unknown_lamer: hmm? no I just grabbed the version from sourceforge <unknown_lamer>he was promoted for it too, even though it may hae led to vlc leaving >:O <unknown_lamer>it sucks really, because sf.net is free software and github is evil SaaS... <dsmith-work>Bah. Just reading the monthly update from SF. In the section about sharing GitHub files on SF, does "take advantage of SourceForge's distribution know-how" mean that ad-ware wrapper stuff? <civodul>probably, that's definitely part of their expertise :-)