Newest at the top
| 2025-02-12 23:35:43 +0100 | tri | (~tri@ool-44c70bcb.dyn.optonline.net) (Ping timeout: 244 seconds) |
| 2025-02-12 23:34:20 +0100 | alfiee | (~alfiee@user/alfiee) alfiee |
| 2025-02-12 23:33:07 +0100 | tromp | (~textual@92-110-219-57.cable.dynamic.v4.ziggo.nl) |
| 2025-02-12 23:31:32 +0100 | tri | (~tri@ool-44c70bcb.dyn.optonline.net) |
| 2025-02-12 23:31:25 +0100 | <tomsmeding> | https://github.com/haskell/cabal/pull/10792 |
| 2025-02-12 23:29:44 +0100 | <tomsmeding> | anyone here want to be pulled in as a requested reviewer on my docs PR? :p |
| 2025-02-12 23:29:42 +0100 | <geekosaur> | (might be in the Cabal library, I don't recall exactly where I found it) |
| 2025-02-12 23:29:21 +0100 | <tomsmeding> | good :p |
| 2025-02-12 23:29:13 +0100 | <geekosaur> | the quote interpretation I suggested doesnt go through a shell, it's a routine inside cabal-install (IIRC) |
| 2025-02-12 23:28:40 +0100 | <tomsmeding> | right, and I also don't see a shell invocation when I trace exec calls on cabal |
| 2025-02-12 23:28:18 +0100 | <geekosaur> | didn't we demonstrate it's not, though? $() didn't work iirc |
| 2025-02-12 23:27:42 +0100 | <tomsmeding> | I don't think "you'll need to compromise on windows anyway" is an argument for putting this through a shell on non-windows |
| 2025-02-12 23:27:09 +0100 | <geekosaur> | which puts limits on what cabal-install or the Cabal library (as "personified" by Setup.hs) can pull off even at best |
| 2025-02-12 23:27:01 +0100 | <tomsmeding> | so you'll have to use whatever syntax the callee is going to parse |
| 2025-02-12 23:26:39 +0100 | <tomsmeding> | because windows has no concept of separate arguments whatsoever in the first place |
| 2025-02-12 23:26:27 +0100 | <tomsmeding> | okay passing separate arguments on windows is a trainwreck anyway |
| 2025-02-12 23:26:02 +0100 | <geekosaur> | basically it's a trainwreck no matter what you do |
| 2025-02-12 23:25:42 +0100 | <geekosaur> | right, but the problem there is there's no exec() on Windows and CreateProcess passes a single string, not a list |
| 2025-02-12 23:25:41 +0100 | <monochrom> | Why is JSON valid Javascript syntax? Because initially someone just wanted to eval() it. |
| 2025-02-12 23:25:05 +0100 | <tomsmeding> | monochrom: what does JSON have to do with shells? :P |
| 2025-02-12 23:24:55 +0100 | <tomsmeding> | if nothing else because various systems have various different shells that do different unexpected things |
| 2025-02-12 23:24:52 +0100 | <monochrom> | How JSON was invented... :) |
| 2025-02-12 23:24:27 +0100 | <tomsmeding> | putting user input through a shell is almost always a mistake |
| 2025-02-12 23:24:08 +0100 | <tomsmeding> | if they're _really_ interpreted by a shell, I'd argue that's a design bug |
| 2025-02-12 23:23:45 +0100 | <tomsmeding> | geekosaur: are the "quotes" interpreted by a shell or by Setup.hs? Telling signs for the _latter_ are that 'single quotes' don't work, and $() also doesn't |
| 2025-02-12 23:23:03 +0100 | <geekosaur> | that said, part of the problem here is there's no guarantee any of this would work on Windows |
| 2025-02-12 23:22:30 +0100 | foul_owl | (~kerry@193.42.0.124) (Ping timeout: 244 seconds) |
| 2025-02-12 23:21:46 +0100 | machinedgod | (~machinedg@d108-173-18-100.abhsia.telus.net) (Ping timeout: 244 seconds) |
| 2025-02-12 23:21:42 +0100 | <tomsmeding> | --ghc-options accepts the quotes halfway down the option |
| 2025-02-12 23:21:33 +0100 | <tomsmeding> | right |
| 2025-02-12 23:21:02 +0100 | <geekosaur> | quotes around the whole option, because it's not trying to be a shell parser |
| 2025-02-12 23:20:42 +0100 | <geekosaur> | IWBNI if it at least did what cabal files allow in a number of places: `--repl-options='"-ghci-script=hi there.ghci"'` (yes, I tested that this still doesn't work) |
| 2025-02-12 23:19:57 +0100 | Vajb | (~Vajb@85-76-36-81-nat.elisa-mobile.fi) |
| 2025-02-12 23:17:03 +0100 | <tomsmeding> | but I think it's good if the docs describe the end result |
| 2025-02-12 23:16:44 +0100 | <tomsmeding> | I don't know what atrocities the thing I pass to --repl-options goes through, but that's not my responsibility |
| 2025-02-12 23:16:27 +0100 | <tomsmeding> | I know how my shell works :p |
| 2025-02-12 23:15:50 +0100 | <tomsmeding> | merijn: $ touch 'hi there.ghci'; cabal repl --repl-options='-ghci-script="hi there.ghci"' # -> ... Warning: ignoring unrecognised input `there.ghci"' |
| 2025-02-12 23:14:59 +0100 | tnt2 | tnt1 |
| 2025-02-12 23:14:59 +0100 | tnt1 | (~Thunderbi@user/tnt1) (Ping timeout: 260 seconds) |
| 2025-02-12 23:14:38 +0100 | <tomsmeding> | euouae: thanks :) |
| 2025-02-12 23:14:35 +0100 | <merijn> | the does not accept quotes seems suspect, as the quotes generally are eaten by the shell, no? |
| 2025-02-12 23:14:25 +0100 | <euouae> | tomsmeding: I say go ahead, I can't review further because I have some stuff to do away from computer now |
| 2025-02-12 23:13:57 +0100 | tnt2 | (~Thunderbi@user/tnt1) tnt1 |
| 2025-02-12 23:13:38 +0100 | <tomsmeding> | github (tm) |
| 2025-02-12 23:13:35 +0100 | <tomsmeding> | ah yes |
| 2025-02-12 23:13:29 +0100 | <merijn> | ah, no the ...was part of the url xD |
| 2025-02-12 23:13:19 +0100 | <tomsmeding> | merijn: doesn't for me |
| 2025-02-12 23:12:59 +0100 | <merijn> | tomsmeding: that shows empty? |
| 2025-02-12 23:12:54 +0100 | <tomsmeding> | euouae: no thanks this is great |
| 2025-02-12 23:12:48 +0100 | <tomsmeding> | --PROG-options has OPTS, so let's use that |