<spauldo>oh, I thought you meant the chinese characters. I didn't know the north stopped using them. <ijp>no, if I did that I'd spend ages trying to decide on the right character <spauldo>Japanese has a couple other alphabets as well, and mix and match them all. I always thought it was odd, but japanese people always thought I was odd so it all evened out. <spauldo>never got to visit korea, unfortunately. <ijp>Japanese is the weirdest writing system, that's pretty much not up for dispute <profan>ah futhark, a messy history there <profan>especially great how they went from a larger alphabet to one that suddenly couldnt represent some of the sounds in their language at one point (at least in sweden's futhark history) <profan>elder -> futhark reduced the size of the alphabet <profan>random interjection, i couldnt help myself when you mentioned futhark <ijp>yeah, I don't think the anglo saxon version shrunk like the scandanavian verison <spauldo>meanwhile, Odin's sitting there complaining that he plucked out his eye for those characters, and now people think they're too good for them. <spauldo>or I guess it would have been Woden at the point. <ijp>I didn't realise that a part of sweden was still using runes as the 1700s <profan>it was in use for a long time alongside the new latin alphabet <ijp>has stis been on lately? <brendyn>ACTION only knows 2500 Chinese characters ;/ <brendyn>spauldo: There are crazy biases at play with peoples understandings of language. Chinese characters suck, and are especially poor for writing Japanese/Korean. The reason you see the alphabet used in Japanese is that you literaly (literaly literarily?) can't write all japanese in Chinese due to inflections <brendyn>So they have a writing system that works for half of most sentences and two other syllabries for the rest. Japanese & Chinese typically believe Chinese characters are indespensible due to various insanitiess -- see John Defrancis Fact and Fantasy. <ijp>If I hadn't rearranged my bookshelf about two weeks ago, I could even grab that book from where I am sitting <ijp>wleslie: I see, well, I'll catch him sooner or later <spauldo>brendyn: I've never heard it put that way, but then again English isn't much better; a good chunk of the letters in the alphabet are pointless, and the spelling conventions were laid down when it was "middle English" and modern words don't sound anything like their spelling anymore. <spauldo>then Japanese must be extradordinarily bad <brendyn>Japanese words often have a Japanese pronunciation, plus 2 or 3 Chinese pronunciations taken from different points in history. <spauldo>there is the advantage that even though Mandarin and Japanese are completely different, when written using chinese characters a lot of the meaning comes through. <brendyn>There is certainly a lot of beauty in them <spauldo>I wish I had tried to take in more of the culture when I was there, but I was too busy nerding it up in my dorm most of the time. I learned Perl over there, though. <brendyn>But I think that's not all that true. most characters are like "has something to do with water and is pronounced like sheep" = ocean <brendyn>And that's an example that actually makes sense and has preserved the pronunciation. I think the general wisdom of linguists is that language is spoken, and writing is basically phonetic representation of spoken language, essentially. Even Chinese characters are mostly just a poor phonetic symbols <spauldo>I dunno. A Japanese guy was the one that told me he could read Korean and Chinese writing because they used mostly the same writing style in Kanji (the Japanese term for the Chinese characters). I don't remember much else he said though 'cause the music was loud and I was drunk. <brendyn>Sure, you can take Japanese sentences and write them with English words too like me wa brendan is, and then all of a sudden you can read Japanese <brendyn>But somehow writing Japanese with English vocabulary items never occurs to us as a brilliant idea. <spauldo>no, I mean he could read stuff written in Mandarin or Korean using chinese characters <spauldo>not everything, but he could get the gist of it <spauldo>whereas Japanese in Romanji is meaningless to me <brendyn>Japanese is honstly written like that <spauldo>granted, spoken Japanese is meaningless to me. It always sounded kind of like Spanish to me. <brendyn>It's like you write "good" where "good" is pronounced ii <brendyn>Essentially, you're not "really" reading Japanese, you're reading Google translate Japanese, except it's 2000 year old official Google Translate that even Japanese use <brendyn>And the funny thing is that they seem to be doing just fine with it. You can learn to read basically any script fluently regardless of design flaws. The hard bit is learning the language it's self <brendyn>spauldo: If you wanted investigate. You could quiz some Japanese people "Do you think Chinese characters are needed, or can we just use romaji or similar", and then ask a Korean. I think it's a good juxtaposition because they are related languages, and yet Korean uses a very neat alphabet. You will get different responses from them. <spauldo>Yeah, ijp was saying the Koreans mostly ditched the Chinese characters. I don't know a lot about Korea. I only know about Japan 'cause I lived there a couple years. <brendyn>It took them time to change though. Chinese characters = intellectual. Simple phonetic script = plebs <spauldo>Japan was kind of like that. Hiragana was the "women's alphabet." Men were expected to learn Kanji. <spauldo>That's no longer the case, of course, but there's a lot of Japanese poetry that's completely in Hiragana because women weren't allowed to write in Kanji. <brendyn>I mean, gender rolls and strict conformism <spauldo>oh, sure, but they had gender equality thrust on them before they were culturally ready for it. <brendyn>Well I'm sorry for rushing their leasurely justice preparations <spauldo>it'd be like if the US took over Saudi Arabia and gave women equal rights. It takes a few generations for people to adapt. <spauldo>yes, although don't think I'm arguing that we should go around taking over countries. <brendyn>I'm being a bit cocky. If you watch enough Noam Chomsky talks, you'll see the kinda perspective that I find easy to believe <spauldo>I'm not saying that giving women rights was wrong, just that it's not the way it normally happens. Usually the laws change within a few decades of public opinion. In Japan's case, they went from no rights to equal rights overnight. <brendyn>And I was reading Clintons emails the other day, just clicking around, reading about how she wants to go and massacre the libya until "libya says it's friends with the US" <spauldo>I've never read or watched any of his stuff, only heard him on NPR a few times. <brendyn>He's like a international politics version of Stallman ;P <spauldo>I get enough Stallman from Stallman :) <spauldo>where I live, I basically have no vote. I'm in a red state. Whoever has an (R) next to their name wins, no matter what. So I don't pay much attention to politics. <spauldo>I'll have an extra large helping of Stallman this semester, actually. I'm taking "Open Source." It's a class on business and legal considerations of open source software from a corporate point of view. <brendyn>I mean in the sense that Stallman is like "How about we just not subjugate everyone?" And Chomsky is like "How about we just not carpet bomb everyone?" <mark_weaver>spauldo: somehow I suspect there will be very little Stallman in that course. <spauldo>In the course material, no. The instructor seems to equate Stallman with all things Open Source, though. <spauldo>And I'm the only one in the class that's been around long enough and worked with Open Source enough to be able to correct him. <mark_weaver>stallman's philosophy bears almost no resemblance to that of "open source" <spauldo>My instructor being incorrect is pretty much par for the course, I'm afraid. <mark_weaver>brendyn: I would say that their perspectives are compatible and that many people agree with both of them, but there are also large numbers of people who would agree with one but not the other, in both directions. <brendyn>I guess I can imagine that, althoug it's too specific of a demographic for me to imagine who would disagree with one but not the other <spauldo>I imagine there's a fairly small cross section of people into both hard political philosophy and open source / free software. <spauldo>I mean, I know Stallman likes to talk about international politics sometimes, but honestly I don't pay attention to anything he says that isn't software related. He's a smart guy, sure, but why would I care about a plumber's opinion on basket weaving? <brendyn>spauldo: Well he maintains a Politics twitter-like feed <brendyn>It's mostly not about politics in the sense of how to run a country but more like "this is bad and this is good" <mark_weaver>spauldo: fwiw, I think Stallman is well worth listening to about international politics. he has more see stallman.org. <spauldo>I've seen a bit of it. But to me, it's not much different than when a movie star endorses a political candidate. <mark_weaver>comparing his opinions on international politics to a plumbers opinion on basket weaving is not a useful analogy here, I think. in fact, the set of skills that he has, and that are needed for his work in free software, overlaps quite a bit with what it takes to have interesting opinions on international politics, imo <spauldo>yes, but here's the problem: Stallman's not evil. <brendyn>spauldo: I think You've made the wrong association. Stallman's achievements are not of that type. They are entirely intellectual or talent based. Building an operating system wrong scratch, starting a foudation, inventating a license that exploded in popularity <spauldo>Sure, but as far as political power goes, he's just another voice among millions. He doesn't have what it takes to work within power structures. He won't budge on something he feels strongly about. <spauldo>yeah, we've been off topic for a while now. My bad. <mark_weaver>you could say the same thing about Martin Luther King, or Gandhi <brendyn>I don't really care about that. It'd be strange If an IRL programming club vehemently banned non programming topics <brendyn>We are here because we are interested in guile, and hence we chat <spauldo>They both had political power. They had millions behind them, and they frightened the people in power. Stallman's a smart guy with an opinion, but he doesn't have millions to storm the streets on his behalf. <spauldo>In the software world, Stallman is a definite somebody. Ask the average politian about Richard Stallman, and they say "Who?" <spauldo>and honestly, time spent reading political opinions is time I could spend procrastinating on my homework. <spauldo>Now, if Stallman could rally enough people to make Guile Emacs a good replacement for GNU Emacs... he'd be my new best friend. <mark_weaver>what's needed there is to do the technical work for making guile emacs work well. <brendyn>I think Guile needs to get stronger and have more users. Some Emacsen aren't confortable with guile when is basically just Andy Wingo - relative to Emacs contributions <spauldo>What's really needed is to redesign Emacs completely with a Scheme (or even Common LISP) base, get rid of the old cruft, and nix elisp all together. Open up the opaque objects to be manipulatable from Scheme code. And I know that will never happen, but I can dream. <mark_weaver>it already exists in an editor called "Edwin", part of MIT/GNU Scheme <brendyn>spauldo: Do you mean even the C parts of Emacs? <spauldo>Sure, but it doesn't have all the features of Emacs. <mark_weaver>and its fate demonstrates the fundamental flaw in that idea :) <spauldo>not all the C parts, no. But one of the common complaints about Emacs is that there's a lot in the C code that would be nice to able to change within elisp. <brendyn>Ok. We'll write it in scheme and then transpile the scheme back the C and compile it <spauldo>mark_weaver: I don't know if I agree with that. Better doesn't always win. There's a lot of other factors that contribute to the success of a project than just how well designed it is. <brendyn>I'm not keen on restarting 30 years of collective effort <spauldo>I know. That's why I said it would never happen. <brendyn>Do you even think it's a good ideda? <spauldo>It's a dream. Would it be nice if I could just snap my fingers and *BAM* thirty years worth of development on a scheme-based extensible editor - with a community to go along with it - would just appear? Yes. <brendyn>You may as well just make the 100 years lisp <spauldo>Oh, here in another thirty years someone would need to snap their fingers to make a new editor - this time with sixty years of development - in whatever the favored version of LISP is at that point. <spauldo>I mostly just want a Scheme-based Emacs because what little I've worked with Scheme, I like it better than elisp. And considering I write more elisp than anything else these days, it'd be nice if I was working in a language that was a bit more clean and supported things like tail recursion. <spauldo>Clojure is all right, but Emacs is heavy enough without throwing the JVM into it. I can justify installing Emacs on a small OpenBSD nameserver, but not the JVM. <brendyn>I think people will still use elisp on guile-emacs <spauldo>yeah. That's what a lot of the problem is, from what I've read - getting an elisp interpreter than can run the entire emacs codebase on top of a scheme-based Emacs engine. <spauldo>it's a solvable problem, it just needs people to work on it. <spauldo>which is why I said RMS would be my new best friend if he got people to get it to the point it could replace emacs <brendyn>(also see the parents of the parent of that <wleslie>spauldo: clojurescript might be easier <spauldo>wleslie: maybe, I haven't messed with it. But then again, it compiles to javascript - and I'm not sure I want an editor based on javascript. That language gives me the hives. <spauldo>not really. Java just makes me sleepy. <spauldo>"wake me up when we get past the boilerplate" <spauldo>besides, clojure doesn't compile to java; it compiles to the JVM <wleslie>the thing is you don't have to interact with it <wleslie>besides maybe some interop to use some native objects you need <wleslie>x86 is awful, and there are lots of decent languages that compile to that <spauldo>it'd make more sense to compile to a vm rather than go through another language. I know clojure itself could, at one point, compile to Microsoft's vm, so how difficult would it be to adapt it to another vm? <spauldo>there's a few vms littered around in the GNU project. <wleslie>spauldo: you said that you couldn't justify installing the JVM on a small openbsd server <spauldo>brendyn: yeah, guile is an example, but I don't know the specifics. How feasible would it be to have another language target the guile vm? <wleslie>v8 or jsc on the other hand are reasonably-sized. <brendyn>spauldo: There is already a elisp and lua now basically and javascript I think? <brendyn>the lua one is not just converting to scheme, but a guile compiler to guile byte code I think <spauldo>well, there you go. All we need to do is rip the vm out of guile, stuff it into a basic editor framework, and write a clojure interpreter that can target it. <wleslie>I had kind of wondered if emacs might be a guile extension rather than embedding guile, that seems to be what the discussion above amounts to <spauldo>no, I agree with the guile-emacs folks; the way they're doing it is really the only feasible way of doing it. Regardless of language, if you can't port over the existing elisp code easily, it's a nonstarter. <spauldo>well, I'm not the best person to make this call, but if I had to design it, I'd probably make a very basic framework in C and make all the primitives available in guile - including threads, some generic window system calls, etc. <spauldo>do 99% of the editor in scheme, using C only for primitives and for places where performance is an absolute must <spauldo>but I'll be the first to admit that I don't really have experience in this area. <wleslie>I like the idea of being able to use buffers and other emacs features in my own programs <wleslie>I do understand that's a way off, though <spauldo>that would be interested - embed emacs in other programs. Usually it winds up going the other way. <spauldo>if I could embed an emacs editor into something like Eclipse - with the various hooks and whatnot that allow Eclipse's functionality to work well with it - it'd make a lot of things easier. Emacs is great, but I can't escape the feeling that I'm standing on top of a giant pile of hacks whenever I try to do something like edit Java code. <spauldo>it's like - "to get C++ working right, you need these 27 addons - some of which are horribly complicated and the documentation is out of date - and you'll still get weird bugs that no one else ever gets." <spauldo>I had to give up on semantic altogether. It caused way more problems than it solved for me, and the codebase is huge and hard to track down where the problems are. <spauldo>and the recommended way to edit Java code is to start up an eclipse server to do all the heavy lifting for you. <catonano>where's the code for hash-fold ? M-. couldn't jump to the definition <spauldo>it's in /libguile/hashtab.c line 1290 <spauldo>unless I'm completely off, it's implemented in C <spauldo>looks like it calls scm_internal_hash_fold, which is defined in the same file on line 1405 <spauldo>hrm, sfri-9 records won't quite do what I want. I need multiple constructors and I need accessors for computed values. <spauldo>or implement the data-directed approach from sicp <amz31>spauldo: what I do usually but I have not a lot of experience doing scheme, is create a record with a constructor that is not public <sneek>Welcome back amz31, you have 1 message. <sneek>amz31, civodul` says: nice post on the graph db! <amz31>spauldo: and add procedure that call the private constructor <amz31>that's what happens in the graphdb context. I have a (make-vertex uid assoc) procedure which create a <vertex> record <amz31>but that procedure is not public, the public interface are (create-vertex assoc) and (get uid) ***amz31 is now known as amz3
<amz3>because there is two ways to have a <vertex> record in the public API <amz3>1) (create-vertex assoc) persist to disk the assoc and create a vertex from it <amz3>2) (get uid) you know a vertex you want to retrieve <amz3>(make-vertex uid assoc) is not useful as public API, especially since `create-vertex` mess with the assoc before saving <amz3>so using make-vertex directly could break the db <rekado>I'm using guile-redis and I wrote a little syntax wrapper to establish a connection with the redis server and always disconnect cleanly. <rekado>Now I wonder if this could be a problem when I'm using multiple processes to connect to redis. <rekado>I guess that's up to the redis server to deal with multiple simultaneous connections. <davexunit>guile-redis, on the other hand, is implemented in a less-than-great way <ijp>I wrote a redis wrapper years back that was portable r6rs, but I later realised it was ass and didn't release it <rekado>for this web interface I'll need quite a lot of worker processes to ensure that submitted jobs are processed without too much of a delay <ijp>the wrapper, not redis <rekado>davexunit: what's wrong with guile-redis? <ijp>my memcached one is solid <ijp>well at least a colloid <rekado>looks like it merely produces strings and sends them over a socket <davexunit>I like the author a lot, I'm glad he made both of these libraries, but I would like them to reimplemented. <ijp>rekado: macro obsession is kind of a phase lispers go through <davexunit>it's been awhile since I used guile-redis, but I couldn't do something that I wanted to do because of the macros <ijp>did redis ever make a binary api? <ijp>memcached has one, which makes writing a wrapper really nice <rekado>davexunit: one thing that doesn't work is renaming or hiding procedures, such as the poorly named "append". <rekado>that's because of the re-export macro magic. <quigonjinn>what's the equivalent of (compile exp #:to 'assembly) in guile 2.1.3? <ijp>if anyone wants to see what I would prefer *instead* of just a sexp version of SQL, you could look at the 500 line "SQL to C" compiler from the ICFP year before last <paroneayea>btw, I think postgres has a language layers structure where you can plug in non-sql languages <paroneayea>I've wondered what would happen if an sexp'y layer was added at that level instead <mark_weaver>ijp: can you elaborate, and maybe provide a link? are you thinking of an SQL to Scheme compiler, or something? I'm not sure I understand. <ijp>paroneayea: there was a guile thing for postgres but I expect it is bitrotten <mark_weaver>I'll have to look at this later. currently working on something time-sensitive. <paroneayea>I'd like to have CREATE replaced with '(create ...) instead <ijp>I have an elaboration on the idea, but I'll wait till the weekend to see if I can't figure out why it's a bad idea