Home freenode/#haskell: Logs Calendar

Logs: freenode/#haskell

←Prev  Next→
Page 1 .. 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 .. 5022
— 502,152 events total
2020-11-23 00:39:21 <dminuoso> If you tell me "this is a haskell compiler", and yet it cant compile most packages dubbed as "haskell package", then why should I buy your implementation?
2020-11-23 00:39:37 <dolio> monochrom: That's probably wrong, too.
2020-11-23 00:39:40 × Tops2 quits (~Tobias@dyndsl-095-033-027-066.ewe-ip-backbone.de) (Read error: Connection reset by peer)
2020-11-23 00:39:46 <dolio> They keep making major language changes.
2020-11-23 00:39:53 <dolio> Or at least, did in the past.
2020-11-23 00:40:03 <monochrom> There is the expectattion that if you called your compiler "purescript compiler" people don't expect it to be compatible with any of Hackage, whereas if you call it "Haskell compiler" people suddenly do.
2020-11-23 00:40:26 <monochrom> Ah OK, oh well.
2020-11-23 00:40:55 <dolio> dminuoso: Why would anyone buy it, period. The problem is, there's nothing that isn't GHC that has a tangible reason for people to use it. The tangible reason to not use GHC comes first.
2020-11-23 00:41:01 × merijn quits (~merijn@83-160-49-249.ip.xs4all.nl) (Ping timeout: 264 seconds)
2020-11-23 00:41:37 <dminuoso> That's a fair point.
2020-11-23 00:41:39 <hekkaidekapus> dolio: You seem to be okay with monocultures.
2020-11-23 00:41:53 <dminuoso> I don't think dolio suggested he's *okay* with the situation.
2020-11-23 00:42:26 <hekkaidekapus> I got the point about implementations preceding standardisation.
2020-11-23 00:42:28 <dolio> I'm describing what the reality is, yes.
2020-11-23 00:42:51 <dolio> If you don't like it, you need to change the reality.
2020-11-23 00:43:10 <stu002> @hekkaidekapus thanks for that
2020-11-23 00:43:10 <lambdabot> Unknown command, try @list
2020-11-23 00:43:28 <hekkaidekapus> stu002: yw
2020-11-23 00:43:59 <dminuoso> Perhaps its actually projects like Frege that could drive alternate implementations.
2020-11-23 00:44:21 × weilzach quits (~weilzach@pool-72-66-93-213.washdc.fios.verizon.net) ()
2020-11-23 00:44:35 <dminuoso> A reason not to use GHC is because you might want more free interop with the JVM if inline-java is not enough.
2020-11-23 00:44:44 <monochrom> I suddenly have a feeling that the Haskell Report died because Paul Hudak died.
2020-11-23 00:44:47 <dminuoso> At the very least that's a large industry.
2020-11-23 00:45:03 <dolio> If there were a real alternative to GHC that did something way better, but didn't implement all of GHC's stuff, there'd be people saying, "hey, can you make stuff compatible with this other thing? Maybe we should standardize it."
2020-11-23 00:45:37 <dminuoso> dolio: Yeah. Frege fits that description exactly I guess.
2020-11-23 00:45:57 <dolio> But making the alternative is the significant work, and just asking people to behave as if that other thing existed won't make the work happen, and eventually people will notice that you're asking them to do a lot of work conforming to your morality for no real benefit.
2020-11-23 00:46:36 → cosimone joins (~cosimone@5.171.24.24)
2020-11-23 00:46:56 <monochrom> Yeah, hell, when I write in C I code to folklore x86-64 C, not the C standard.
2020-11-23 00:47:32 <dminuoso> Heh, barely anyone ever adheres to the C standard
2020-11-23 00:47:36 × cr3 quits (~cr3@192-222-143-195.qc.cable.ebox.net) (Ping timeout: 240 seconds)
2020-11-23 00:47:57 <monochrom> When I teach Haskell to students I shamelessly say "lazy evaluation" not what the Haskell Report tries to avoid to say.
2020-11-23 00:48:16 <monochrom> (BTW IMO the Haskell Report also fails in its attempt to shy away from lazy evaluation.)
2020-11-23 00:48:22 <dminuoso> Most of C appears to be not even folklore C but "whatever I imagine"-C
2020-11-23 00:48:29 <hekkaidekapus> Standards are for compiler implementors anyway.
2020-11-23 00:48:42 <monochrom> (For example explain to me why else "seq" is called "seq" not merely "strict" or "strictify")
2020-11-23 00:48:45 <dminuoso> hekkaidekapus: That's a pretty inadequate bar, honestly.
2020-11-23 00:48:52 <hekkaidekapus> heh
2020-11-23 00:49:00 <dminuoso> If only language implementors can read it, how should a language user understand your language?
2020-11-23 00:49:45 × cosimone quits (~cosimone@5.171.24.24) (Client Quit)
2020-11-23 00:50:11 <dminuoso> If you take this to the extreme you end with something like C++ where not even the C++ committee understands what some of their standard means.
2020-11-23 00:50:22 → cosimone joins (~user@5.171.24.24)
2020-11-23 00:50:35 × cosimone quits (~user@5.171.24.24) (Remote host closed the connection)
2020-11-23 00:50:56 × conal quits (~conal@66.115.157.141) (Quit: Computer has gone to sleep.)
2020-11-23 00:51:09 <dminuoso> [13] is a great source for inconclusive debates
2020-11-23 00:51:32 <hekkaidekapus> I bet LLVM and GCC coders have a great grasp of the standard (those not on the C++ committee).
2020-11-23 00:51:42 → cosimone joins (~user@5.171.24.24)
2020-11-23 00:51:45 hekkaidekapus is just hyperboling at this point.
2020-11-23 00:52:02 <int-e> monochrom: you like puns, how would you like an independent Haskell implementation without type literals and type-in-type shenanigans? :)
2020-11-23 00:52:25 <monochrom> Hrm, why is pun involved in that?
2020-11-23 00:52:34 × cosimone quits (~user@5.171.24.24) (Remote host closed the connection)
2020-11-23 00:52:50 <monochrom> Oh hahahahaha
2020-11-23 00:52:59 <int-e> (More seriously though, without another relevant Haskell implementation, the point of a standard is rather moon.)
2020-11-23 00:53:06 <int-e> moot.
2020-11-23 00:53:26 <hekkaidekapus> Define: relevant.
2020-11-23 00:53:40 <hekkaidekapus> Implementing the previous standard?
2020-11-23 00:53:50 <int-e> No, actually in use.
2020-11-23 00:54:07 <monochrom> I have a weak reason for wishing for a standard even though there is only one implementation. Much more precise, reliable documentation of what the hell the implementation does.
2020-11-23 00:54:13 × phaul quits (~phaul@ruby/staff/phaul) (Ping timeout: 264 seconds)
2020-11-23 00:54:17 → conal joins (~conal@66.115.157.141)
2020-11-23 00:54:36 <dolio> Weren't you the one that said GHC is better documented than the actual standard? :þ
2020-11-23 00:54:41 <dminuoso> int-e: okay it took me a minute to get the pub. quite cute
2020-11-23 00:55:26 → cr3 joins (~cr3@192-222-143-195.qc.cable.ebox.net)
2020-11-23 00:55:54 <monochrom> I think I didn't say quite that quite directly. What I really think though is GHC is less worse than the Haskell Report.
2020-11-23 00:56:34 <monochrom> And GHC's is somewhat better than "less worse" (but not a lot better) if you also include the papers it cites.
2020-11-23 00:57:19 <monochrom> For example if you're looking for the typing rules of GHC 9's RankNTypes, you'd better read that bidirectional inference paper.
2020-11-23 00:57:37 <dolio> Yeah.
2020-11-23 00:57:39 <monochrom> rules that should have been in a standardized document.
2020-11-23 00:58:18 → Jonkimi727406120 joins (~Jonkimi@113.87.161.66)
2020-11-23 00:58:18 × chkno quits (~chkno@75-7-2-127.lightspeed.sntcca.sbcglobal.net) (Read error: Connection reset by peer)
2020-11-23 00:58:33 <dolio> It'd be nice if their wiki gave a compilation of how it works, since you never know what all from the past is relevant.
2020-11-23 00:58:34 → chkno joins (~chkno@75-7-2-127.lightspeed.sntcca.sbcglobal.net)
2020-11-23 00:58:36 → cosimone joins (~user@5.171.24.24)
2020-11-23 00:59:15 × gehmehgeh quits (~ircuser1@gateway/tor-sasl/gehmehgeh) (Quit: Leaving)
2020-11-23 00:59:29 <Axman6> The top story on lobste.rs has the title "Booting from a vinyl record" and I was sad to see it wasn't haskell related
2020-11-23 00:59:51 <monochrom> heh
2020-11-23 01:00:55 × cosimone quits (~user@5.171.24.24) (Client Quit)
2020-11-23 01:02:36 <monochrom> err, the bidirectional inference paper for pre-9 RankNTypes; the quick-look paper for post-9 RankNTypes and impredicativity.
2020-11-23 01:03:04 <monochrom> but the latter tells you that they assume you have read the former, so you're still toasted.
2020-11-23 01:04:39 <monochrom> What I fear is that if a comittee were formed to add these to Haskell Report, a very vocal, counterproductive minority would argue "cannot go with these rules verbatim, we need to allow different compilers to make different trade-offs"
2020-11-23 01:04:59 × LKoen quits (~LKoen@169.244.88.92.rev.sfr.net) (Quit: “It’s only logical. First you learn to talk, then you learn to think. Too bad it’s not the other way round.”)
2020-11-23 01:05:14 <monochrom> And that will land you with: either no change to the Haskell Report, or the C standard.
2020-11-23 01:07:03 <dolio> Yeah, that would probably be useless.
2020-11-23 01:07:18 × tv- quits (~tv@unaffiliated/tv-) (Quit: WeeChat 2.8)
2020-11-23 01:08:08 <dolio> Although what would the situation be if it were standardized? GHC keeps changing what you're allowed to do.
2020-11-23 01:08:15 → olligobber joins (olligobber@gateway/vpn/privateinternetaccess/olligobber)
2020-11-23 01:08:20 <dolio> Would the standard keep shifting?
2020-11-23 01:10:59 → vicfred joins (~vicfred@unaffiliated/vicfred)
2020-11-23 01:11:09 <monochrom> That's a tough one. My preference is GHC changes quickly, the standard changes slowly to summarize the sweet spots GHC has found.
2020-11-23 01:11:14 <dolio> I think a lot of the development has been allowing it to infer strictly more cases, but sometimes it has switched trade-offs.
2020-11-23 01:11:47 × whataday quits (~xxx@2400:8902::f03c:92ff:fe60:98d8) (Remote host closed the connection)
2020-11-23 01:12:54 → whataday joins (~xxx@2400:8902::f03c:92ff:fe60:98d8)
2020-11-23 01:13:30 <dolio> Oh, and of course, keeping multiple, distinct checking algorithms around in a single implementation wouldn't be very nice. :)
2020-11-23 01:13:41 <dolio> So that wouldn't be a solution.
2020-11-23 01:14:01 × vhs quits (~vhs@41.44.147.255) (Quit: Leaving)
2020-11-23 01:14:07 → guest1123 joins (~user@49.5.6.87)
2020-11-23 01:14:30 × conal quits (~conal@66.115.157.141) (Quit: Computer has gone to sleep.)
2020-11-23 01:15:32 → merijn joins (~merijn@83-160-49-249.ip.xs4all.nl)
2020-11-23 01:16:31 <monochrom> Windows 95 managed to have two kernels inside it haha
2020-11-23 01:18:14 <monochrom> to be sure, you would be fair to point out "that exactly shows what wrong with it"
2020-11-23 01:18:46 × Feuermagier quits (~Feuermagi@213.178.26.41) (Remote host closed the connection)
←Prev  Next→
Page 1 .. 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 .. 5022

All times are in UTC.