Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2017-02-10

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

All times shown according to UTC.

Time Nick Message
02:47 ilbot3 joined #darcs
02:47 Topic for #darcs is now http://darcs.net/ | logs: http://irclog.perlgeek.de/darcs/ | darcs 2.12.5 is out http://darcs.net/Releases/2.12
04:06 mizu_no_oto joined #darcs
06:50 mornfall joined #darcs
12:18 mizu_no_oto joined #darcs
12:40 mizu_no_oto joined #darcs
13:57 mornfall Heffalump: did you ever get to the conflict marker ordering/annotation problem?
14:33 Weltraumschaf joined #darcs
14:40 byorgey joined #darcs
14:40 byorgey joined #darcs
15:54 peb` joined #darcs
17:59 bluepixel joined #darcs
18:12 pem_ joined #darcs
18:23 bluepixel hey. is there a way to concatenate darcs cmds like git aliases yet?
18:29 Heffalump mornfall: not any further than my experiments from years ago
18:29 Heffalump bluepixel: nope (I'd never actually heard of that as a feature request, do you have some examples you'd like to do?)
18:30 Heffalump mornfall: those experiments showed that I could calculate the source patches for hunks in a conflict, but it was expensive to do on the fly so really needed a repo format change to capture the information in patches.
18:31 bluepixel Heffalump: e.h. record and push in one command. the problem is i can't use shell aliases due no parameter allowed.
18:31 bluepixel e.g.*
18:37 mornfall Heffalump: too bad, I suppose that means there isn't a darcs branch I could download and get something (slow wouldn't bother me much)
18:41 Heffalump mornfall: it exists somewhere, let me try to find it
18:42 Heffalump http://hub.darcs.net/ganesh/darcs-conflict-marking
18:43 Heffalump no idea how far back in GHC-land you'll need to go to build it though
18:49 Riastradh joined #darcs
18:57 mornfall thanks, will try (my ghc is old-ish all by itself, so maybe it'll work out for the best)
18:57 mornfall ok, cabal barfed a bunch of conflicts and died
18:58 mornfall next try when I get home
19:52 mornfall I suppose I am a bug magnet... this might be more work than it'd save :\
20:06 mornfall yeah of course haskell haskell platform would be dynlinked to libraries that are gone from 'modern' linux
20:09 drostie_ joined #darcs
20:11 mornfall somewhat regrettably, ghc won't build if the existing ghc is newer than the one you are trying to build
20:12 sm bluepixel: you can do parameters with shell aliases, or shell functions
20:13 mornfall just for the kicks, tried darcs pull --skip-conflicts from screened and got conflicts (darcs 2.10)
20:13 mornfall I probably shouldn't touch anything at this point
20:14 sm building that darcs-conflict-marking branch looks like an adventure - in 2011, ghc 7.2 was current
20:16 mornfall sm: I have a whole lot of options there, a rhel6 box without cabal-install but with ghc 7, arch linux with ghc 8 but without libncurses.5 (ghc 7.2 fails to build), openbsd 6 box with ghc 7.10 (ghc 7.2 fails to build)
20:16 mornfall the rhel6 box is going to get upgraded to rhel7 tomorrow though, so I guess my most viable option is a very short-lived one
20:17 mornfall but maybe rhel7 will still have libncurses.5 ... which won't help in the end, because haskell platform can only be installed in /usr/local/haskell (and I am not root on that one box)
20:17 sm sounds like days of fun :)
20:18 mornfall sounds like maybe I could write this in perl from scratch and still save time
20:18 sm maybe the changes could be ported forward to modern darcs
20:19 mornfall sm: well, the number of patches in my original problem (where I could use better conflict markers) and in darcs screened vs the branch happens to be about equal
20:19 mornfall time saved, minus a lot
20:20 sm I wouldn't bother about haskell platform
20:21 sm I'd install ghc 7.2, and if that doesn't work on my machines I'd spin up a clean vps where it does work
20:21 sm and make a note to fix my machines
20:22 mornfall ghc won't build if your existing ghc is newer than the one you are trying to build
20:22 sm oh. Um.. might have to set up that vps with an old debian version then
20:24 sm but, you can install the ghc binaries, surely
20:25 mornfall yeah but I don't have a linux box that's old enough
20:26 sm what happens if you try to install https://www.haskell.org/ghc/download_ghc_7_2_2#x86linux eg ?
20:26 mornfall well, it could be kludged somehow, but I am out of time
20:29 mornfall checking for path to top of build tree... utils/ghc-pwd/dist-install/build/tmp/ghc-pwd: error while loading shared libraries: libgmp.so.3: cannot open shared object file: No such file or directory
20:32 sm I remember that one.. use system package manager to install a newer/older libgmp
20:32 mornfall there's only one libgmp on arch, AFAIK
20:33 mornfall anyway, this is a waste of time
20:33 sm agreed
20:38 Weltraumschaf joined #darcs
20:51 Heffalump mornfall: don't you use nixos? You could just go back in time in nixpkgs to get a working GHC
20:52 Heffalump maybe it's my imagination that you use it
20:52 mornfall Heffalump: no, that suffered from the same problems ghc does (excessive backward incompatibility with itself)
20:53 mornfall yes, with nixos that'd be an option, presumably (only ghc from that era would have long been garbage collected)
20:54 mornfall and building nixpkgs from source is no better than building 2011 darcs from source :\ probably worse, because half the source tarballs would no longer be around
20:55 mornfall anyway, my primary system is now an old-school unix (openbsd)
20:58 Heffalump ah, I haven't tried going back more than a couple of years with it
20:58 Heffalump I've always thought nixpkgs would be a great use case for darcs (if it scaled well enough) - huge numbers of non-conflicting patches you'd often like to reorder
20:59 Heffalump when you say backward incompatibility, do you mean that latest nix can't process old nixpkgs?
21:01 mornfall I more meant that nixos (and nixpkgs, to a lesser degree) configuration periodically changes, so you can't go back and forth without also rewriting your config files
21:02 mornfall upgrading nixos was becoming a multi-day task by the time I gave up on it
21:06 Heffalump oh, right - but you don't really need config files to use an old nixpkgs
21:07 mornfall no, you don't (and vice versa, you can get one thing from new nixpkgs even if your system is stuck with a version from 6 months ago and you can't find time to migrate it to the new config format)
21:07 Heffalump right
21:07 Heffalump most of my development now is done with frozen nixpkgs snapshots
21:08 mornfall on the downside, nixpkgs ghc used to be pretty awful (maybe it's better now)
21:08 Heffalump that said, one of those snapshots has rotted somehow (not sure if I screwed up or somehow the build isn't pure)
21:08 mornfall awful as in, cabal flopping out all the time because registration failed because something something (probably tried to write into the store)
21:09 mornfall and of course even if that is fixed, this would come back with old nixpkgs :-)
21:09 Heffalump :-)
21:09 Heffalump I didn't ever have any problem with that even in the old(ish) Haskell world with ghcWithPackages
21:09 Heffalump also,modern cabal new-buidl will work with GHC 7.0
21:10 Heffalump and that is highly unlikely to try to use the /nix store
21:11 mornfall life's a misery
21:26 bluepixel sm: nice. i didn't know i can embed a function in an alias. :) thanks
22:13 leg joined #darcs

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