Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2016-03-14

| Channels | #darcs index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:15 lambdabot joined #darcs
02:12 mizu_no_oto joined #darcs
02:32 fr33domlover joined #darcs
02:47 ilbot3 joined #darcs
02:47 Topic for #darcs is now http://darcs.net/ | logs: http://irclog.perlgeek.de/darcs/ | darcs 2.10.3 is out http://darcs.net/Releases/2.10
02:59 castlelore joined #darcs
03:12 pointfree joined #darcs
03:35 mizu_no_oto joined #darcs
04:35 theone- joined #darcs
08:02 fr33domlover Heffalump, I decided to cancel the MonadIO thing, I realized nothing I did requires it. Maybe it's not that useful after all. I hope it's possible to delete a patch after I made several on top :P advice welcome, i'm going to go over darcs commands and pray
08:14 fr33domlover hmmm so there's no safe way?
08:16 fr33domlover sm, Heffalump - is it okay if I cancel the MonadIO patch using 'darcs rollback'? or should I create a fresh repo and re-apply all the patches except the MonadIO one?
08:16 fr33domlover i suppose it depends on (1) how easy it is (2) how much Heffalump cares about perfect history in the 'ssh' repo
08:20 fr33domlover hmmm even rollback won't work, it refuses to cancel a patch without the ones following it
09:01 fr33domlover YAY forget everything I said
09:02 fr33domlover MonadIO /does/ have a use case: all request handling is done within a single thread!
09:02 fr33domlover it means you can do it all inside a single monadic computation
09:02 fr33domlover only thanks to MonadIO here
09:02 * fr33domlover wipes the sweat
09:19 haasn joined #darcs
11:02 Heffalump :-)
11:02 Heffalump otherwise you could have used rebase to remove that patch and fix up any conflicts in the following ones
11:26 mizu_no_oto joined #darcs
11:40 gal_bolle joined #darcs
12:20 fr33domlover Heffalump, I was using darcs 2.8 so some of the cool new stuff was missing. But today I stack-installed darcs 2.10.3
12:20 fr33domlover btw almost ready for a merge
12:21 fr33domlover i'm making the monad transformers into newtype wrappers and adding export lines to all modules, that should hopefully be the last API-breaking change
12:22 fr33domlover then mabe i should merge pointfree's changes before I continue to replace OpenSSL with tls?
14:14 Big_G joined #darcs
16:23 gh_ joined #darcs
16:25 pointfree joined #darcs
17:30 amgarching joined #darcs
17:40 theone- joined #darcs
17:48 Riastradh joined #darcs
17:59 amgarching joined #darcs
18:26 amgarching joined #darcs
18:43 owst joined #darcs
18:57 Heffalump fr33domlover: ask pointfree what the status is, I'm not sure they're close to mergeable yet
20:15 sm go fr33domlover !
21:18 pointfree Yeah, the curve25519-sha256@libssh.org kex support is not ready yet. Today I'm making sure the bytes are being laid down correctly in the kex conversation and in the exchange hash structure. I'm stepping through (haskell) ssh server debug output set to True and openssh7 with wireshark. Then the haskell ssh server has to support the old diffie-hellman-group1-sha1 in addition to the newer kex for transitional purposes I would think (o
21:19 pointfree http://hub.darcs.net/ganesh/ssh/issue/5#comment-20160314T043642
21:23 fr33domlover pointfree, are you adding a new dependency for that? i need a crypto-secure random number gen and 'ssh' already uses DSA and RSA so maybe we can just move to using cryptonite for all of this?
21:24 fr33domlover although cryptonite cheats by being written in C right under the Haskell wrappers...
21:31 pointfree I'm using "System.Random" from "random" presently...
21:33 pointfree I'm adding https://hackage.haskell.org/package/curve25519 as a dependency.
21:34 pointfree checking-out cryptonite...
21:37 pointfree Happens to have this too: https://hackage.haskell.org/package/cryptonite-0.13/docs/Crypto-PubKey-Curve25519.html
21:38 dolio System.Random probably has the performance of a cryptographically secure rng without the security. :)
21:40 pointfree "haskell-libssh - A Haskell binding to the server part of libssh." https://github.com/lpeterse/haskell-libssh
21:40 pointfree Just a thought.
21:59 fr33domlover pointfree, I'm picking a random number gen right now too. Their APIs are all very similar so when you and I have closer-to-ready code, let's pick one together and use it :P for now use whatever you like, i suppose
21:59 pointfree I have no strong preferences right now.
22:13 fr33domlover Heffalump, a thought of mine: Right now 'ssh' uses the following: 'HsOpenSSL' for DSA and RNG; 'RSA' for RSA, 'random' for RNG too (oh no, sounds bad) and now we'll have curve25519 too. Ah, and 'SimpleAES' for AES. What if instead of picking separate packages we use 'cryptonite' for all of that?
22:39 Heffalump TBH I don't know enough to have a strong opinion, but it sounds sensible
22:57 sm sounds awesome
22:58 sm cryptonite! take that, crackers!
23:33 fr33domlover i'll try that later, for now i just want to replace HsOpenSSL with minimal changes. pointfree, is the keyxDHInit function relevant or is it part of the old crypto you want to replace/deprecate?

| Channels | #darcs index | Today | | Search | Google Search | Plain-Text | summary