racket: patching out runtime variant detection, fix #114993 #118490
Conversation
|
Result of 2 packages built successfully:
15 suggestions:
Result of 1 package failed to build:2 packages built successfully:
15 suggestions:
Note that build failures may predate this PR, and could be nondeterministic or hardware dependent. |
| @@ -71,6 +71,7 @@ stdenv.mkDerivation rec { | |||
| buildInputs = [ fontconfig libffi libtool sqlite gsettings-desktop-schemas gtk3 ] | |||
| ++ lib.optionals stdenv.isDarwin [ libiconv CoreFoundation ncurses ]; | |||
|
|
|||
| patches = [ ./variant.patch ]; | |||
There was a problem hiding this comment.
Can you add a comment why we need this?
|
I am sorry, I don't have time to review this right now, but I'd like to link racket/racket#3685 which seems to be relevant. In any case, thank you for your work in improving our Racket support! |
|
Result of 1 package marked as broken and skipped:
1 package failed to build:
|
|
It's not directly related, but, for anyone reading this thread, there's been some coordination going on for Nix- and Guix-related issues with Racket over at racket/racket#3707. Among other things, most (hopefully all!) patches to the Racket source should no longer be needed for the forthcoming Racket 8.1. |
| @@ -0,0 +1,54 @@ | |||
| diff --git racket/collects/setup/variant.rkt racket/collects/setup/variant.rkt | |||
There was a problem hiding this comment.
does the following patch work for you?
--- racket/collects/setup/variant.rkt
+++ racket/collects/setup/variant.rkt
@@ -6,7 +6,8 @@
(provide variant-suffix)
-(define plain-variant
+(define plain-variant 'cs)
+#;(define plain-variant
(delay/sync
(cond
[(cross-installation?)
There was a problem hiding this comment.
I think I could, this should save some patch lines :)
Will check results later when build finish
|
ping @corpix |
|
I marked this as stale due to inactivity. → More info |
|
Not relevant anymore |
|
I'm sorry for the late response. |
Motivation for this change
raco exe some-racket-file.rktis not working since update to Racket 8.0, see #114993It fails because Racket try to detect a runtime variant reading a string from it's own executable. Of course it fails to correctly detect runtime type using this method because racket executable is a wrapper script.
This pull request removes patching out a runtime detection mechanism, hardcoding
'csruntime variant (because this is a variant built by racket derivation).I don't like this solution. But maybe someone who will see this pull request could suggest a better way to mitigate the problem?
Things done
sandboxinnix.confon non-NixOS linux)nix-shell -p nixpkgs-review --run "nixpkgs-review wip"./result/bin/)nix path-info -Sbefore and after)