IRC channel logs
2016-08-07.log
back to list of logs
<davexunit>ijp: bit late to the party, but I can answer some questions <Amynka>is there special reason why guile-2.0.12 is not linking libguile.so -> libguile-2.0.so ? <wingo>Amynka: yeah, let me find a link <wingo>libguile-2.0.so is one abi-compatible series <wingo>libguile-2.2.so will be another one <Amynka>wingo: is it I find out that those abi are not compatible :D <Amynka>wingo: so you are basically saying i have to remove it from gentoo and do it all over again :D <wingo>Amynka: hmm i don't know what the implication is for you -- apologies if there was a misunderstanding :) <wingo>the intention is that guile 1.8, 2.0, etc are parallel-installable <wingo>because their library names are different, their headers go in different places, their pkg-config file names are different <wingo>the only collisions between the packages are the "guile" binary itself and the manual <wingo>both of which i think you can uniquify via --program-suffix i think <wingo>also the m4 file, but guile 2.0's m4 file should also work for guile 1.8 <Amynka>yea we will not go for both installed anyway <Amynka>guile 1.8 is super badly written <wingo>a polite word for really old ;) <wingo>doing a flag day from 1.8 to 2.0 is kinda hard tho <Amynka>wingo: you mean like nothing is working with it :)) and about parallel abi i would say in theory yes :)) <Amynka>wingo: but yea I think you did great job on guile-2.0.12 its big progress I like it <wingo>it could be a parallel build failure -- the .i file is generated by the build, maybe we are missing a dep or so <Amynka>wingo: it is totally parallel build <wingo>but -j3, that's not very many, strange to see it <Amynka>yes indeed I tested it on 4 machines including arm with parallel and it was working so.. I have no idea <wingo>i can see how it can happen actually <Amynka>wingo: would be grateful for tips :) need to go off i will read it after thanks :)) <brendyn>I tried compiling guile emacs, and make seemed to spam all the source code out filling up 16GiB of ram <wingo>guile doesn't have an SRE module bundled AFAIU, on the other hand there is one out there somewhere <brendyn>Can we frankenstein scsh into guile :p? <brendyn>it seems most scheme repos are "last edited 3 years ago" <wingo>this one was last edited a couple weeks ago afaiu <brendyn>btw I read your comments on delimited contintuations but I don't really understand them. When would I find myself using them? <wingo>but yeah, given that scheme is old, the average project is not active, and that's very normal i think :) <brendyn>Actually, scsh seems to have a secret repo on github with 0.7 and all. The websites have been neglected. <wingo>TIL debian compiles guile with -O0 <Amynka>wingo: yeey ! :) Thanks :) I will commit it to gentoo and ask him to test it <wingo>probably we should make a new 2.0 release soonish because of some mingw fixes <wingo>maybe it will be the last 2.0 release <Amynka>wingo: lets wait for our users to find more bugs :)) <Amynka>few weeks should be enough apparently whole gentoo upgraded guile. plus I would wanna get rid of 1.8.8 :)) <ArneBab>brendyn: can you throw data recovery tools at it? <brendyn>ArneBab: It does not even come up in fdisk -l. I had to edit it out of /etc/fstab to even boot. Fortunately, I found my imporant files in ~/.saves due to emacs :) and other files of importance were still sitting on my VPS in Japan <ArneBab>that reminds me again that I have to connect my new backup disks… <brendyn>It's quite bothersome, my hard drives are always full since I can't really afford any more so I just have to decide what gets deleted <ArneBab>brendyn: can’t you use a non-ssd disk as longterm storage <brendyn>I do. I have a 1TiB drive and that's also full because I've downloaded so much language learning material I don't want to delete <ArneBab>I recently bought three 3TiB disks (though only two of these are for home usage) <brendyn>I have a 2 TiB which is corrupt, and another 2TiB drive containing the dd_rescue dump of the corrupt drive. It's been sitting there for 5 years with important data I can't figure out how to ever get back <ArneBab>I guess you already tried the common recovery tools? <brendyn>I have a "no-backup" directory containing all the things selected to be lost when this new drive fails, since my only remaining drive is 1TiB and so it can't back up an entire 2TiB <ArneBab>can’t you just buy another drive? They are only about 130€ over here… <brendyn>Not sure really, i didn't try too much. I mounted it as a loop device and then used fsck which told me the partition table was dead or non existant <brendyn>I suppose I can, but I may end up broke soon so I'm being conservative. <brendyn>When I looked online, I think the best tool available was proprietary <ArneBab>(includes guides for several free tools) <brendyn>Suppose I might not be able to recover the file system, just blocks of files <brendyn>It's just going to take 2 days to download 20GiB from Japan :P <paroneayea>wingo: just fyi, I have an urgent deadline of tuesday for my standards group. I'll try to work on the tests before then, but if not, I'll work on them right after. <shymega>paroneayea, how's that tree cutting going? :) <amz3>I see there is not .scm extensions on the web, most people seem to be using .com, .net, .xxx <amz3>did anybody consider submiting .scm as an extensions for the web <amz3>too bad nalaginrut is not here to explain how it works <ijp>registering a tld is pretty expensive <amz3>you have to pay?! it's not a vote or something <ijp>"The evaluation fee is estimated at US$185,000. Applicants will be required to pay a US$5,000 deposit fee per requested application slot when registering." <ijp>most of the new tlds are stupid too <ijp>three different photography ones for a start <amz3>I wonder about .abc it was rumored that google wanted it <amz3>there's .yeah! I just saw it in tv ad <paroneayea>so I wonder what happens when a bunch of these TLDs realize they blew a ton of money they can't recoup <paroneayea>and entire trees starting from GTLDs on down of domains disappear <random-nick>paroneayea: actually you don't have to pay upkeep and can't refund <paroneayea>random-nick: really? I thought it was many many thousands to keep upkeep <paroneayea>5.7 Are there any ongoing fees once a gTLD is approved by ICANN? <paroneayea>Yes. Once an application has successfully passed all the evaluation steps, the applicant is required to sign a New gTLD Agreement (also called Registry Agreement) with ICANN. Under the agreement, there are two fees: (a) a fixed fee of US$6,250 per calendar quarter; (b) and a transaction fee of US$0.25. The latter does not apply until and unless more than 50,000 transactions have occurred in the TLD during any calendar quarter or any <paroneayea>four calendar quarter period. Please refer to section 6.1 of the New gTLD Agreement in the Applicant Guidebook. <paroneayea>all we need is about $200k to start with and $25k from there on out <amz3>i can find 25k guile user and ask them 1$ each <paroneayea>if we're going to raise that much money, maybe we could try to use it to improve guile in some other more direct way :) <amz3>I know my idea is silly, it was funny thinking about it <amz3>people seem to have fun with tlds, I don't understand yet what it means. Maybe it's just a ladder where people spend money to rate themself? <ijp>paroneayea: it's not even just if they gtld goes defunct, some of them have stupid renewal policies <ijp>I think one of the wedding ones only lets you have it for at most two years <ijp>so much for "Cool URIs don't change" <ijp>if you were going to do that, you may as well register .scheme <ijp>but why bother? no other language does <ijp>I wouldn't count an accidental collsion wiht a ccTLD <ijp>a better use of your money would be spending $18000 to be a full member of the unicode consortium <ijp>you can finally get a vote on the elvish codepoints you've been dying for <ijp>(actually tengwar does have some space allocated for it, it just hsan't been officialised) <ijp>(no such luck for klingon though) <amz3>is a monad a procedure applied to another procedure? <ijp>it's an interface that some types implement <amz3>it will remain, a theory for me :) <amz3>I'll rename my search engine project to hyper instead <amz3>I'll rename my search engine project to *culturia* instead <emyles>hi, how can I capture the output of (system "echo $PATH") in a variable? <ijp>well, the better way is to use getenv <emyles>Thanks but I was trying to simplify and it backfired. What I really want to do is get the path to my python, like this: (system "python -c \\"import os,sys; print(os.path.realpath(sys.executable))\\"")) <ijp>emyles: I guess using open-output-pipe from (use-modules (ice-9 popen)) is simplest <emyles>This is for a guix package btw, I think I'll have either to pass -DPYTHON_EXECUTABLE= or patch the package's custom FindPython.cmake <ijp>although ice-9 popen is kinda weird <emyles>okay, thanks I'll find out about pipes <ijp>open-input-pipe not open-output-pipe <wingo>yeah we need some other kind of interface there, the whole "either input or output or input and output" thing is not good <ijp>it's nontrivial to get stderr too <ijp>it's just hard to care when you use it once every other year <wingo>just needs someone to fix it properly :) <wingo>mark_weaver: you have time to review a 2-line fix for #22667? <wingo>maybe i just commit and we revert if it's turrible <wingo>iconv has the wonkiest interface of any standard library function, but i don't know of one better