--- Begin Message ---
Subject: |
[PATCH] gnu: gerbil: Upgrade to 0.17.0. |
Date: |
Sun, 19 Mar 2023 23:06:42 +0000 |
Hello,
Attached is a patch to upgrade Gerbil to the latest version. In addition, it
fixes a few issues:
* switch build code to use g-expressions
* use the `#:tests?' argument of `arguments' instead of deleting the `check'
phase
* run `guix style'
* fix issue with missing version file during build
Honestly, I'm not sure I fixed the latter issue in the most idiomatic way. I
simply printed the appropriate s-expression to the appropriate file using
Guile's `format'. If anyone has advice for a cleaner/more robust way to handle
that, please let me know.
Note that I initially tried to fix the version file issue by ignoring the build
instructions in the Gerbil guide and instead using the autotools build
infrastructure whose artifacts can be seen in the top-level directory of the
Gerbil repository. This ran into a different issue, and in the process of
diagnosing it, I realized that the build infrastructure in question was written
specifically for the previous version. There being no good reason to solve that
problem over this one, and realizing that proper versioning would complicate
the use of that system even more than solving the missing version file issue, I
elected to pursue the route I did.
This is far too much information for a git commit message, but I did feel it
worth mentioning somewhere.
Thanks,
Juli
0001-gnu-gerbil-Upgrade-to-0.17.0.patch
Description: Text Data
--- End Message ---
--- Begin Message ---
Subject: |
Re: bug#62282: [PATCH] gnu: gerbil: Upgrade to 0.17.0. |
Date: |
Sun, 09 Apr 2023 00:01:38 +0200 |
User-agent: |
Gnus/5.13 (Gnus v5.13) Emacs/28.2 (gnu/linux) |
Hello,
Thanks for taking the time to split the patch. I’ve applied #1 and #2.
"J. Sims" <jtsims@protonmail.com> skribis:
> From 569fabf7029200cbb3166af83b8fc89a6378468d Mon Sep 17 00:00:00 2001
> From: Juliana Sims <jtsims@protonmail.com>
> Date: Tue, 28 Mar 2023 23:56:07 -0400
> Subject: [PATCH 3/3] gnu: gerbil: Enable all features.
>
> * gnu/packages/scheme.scm (gerbil): Enable all features.
[...]
> + "--enable-leveldb"
> + "--enable-libxml"
> + "--enable-libyaml"
> + "--enable-lmdb"
> + "--enable-mysql"
> + "--enable-sqlite"
> + "--enable-zlib")))
> (add-before 'patch-generated-file-shebangs 'fix-gxi-shebangs
> (lambda _
> ;; Some .ss files refer to gxi using /usr/bin/env gxi
> @@ -1202,8 +1211,18 @@ (define-public gerbil
> (copy-recursively "../bin" bin)
> (copy-recursively "../lib" lib)))))
> #:tests? #f))
> - (native-inputs (list coreutils gambit-c util-linux))
> - (propagated-inputs (list gambit-c openssl sqlite zlib))
> + (native-inputs (list coreutils
> + gambit-c
> + util-linux))
> + (propagated-inputs (list gambit-c
> + leveldb
> + libxml2
> + libyaml
> + lmdb
> + mysql
> + openssl
> + sqlite
> + zlib))
I have two questions:
1. What’s the impact on ‘guix size gerbil’? Overall we should find a
balance between feature usefulness and size.
2. Do all these really need to be propagated? Shouldn’t they be
regular inputs? If propagation is required, please add a comment
explaining why.
I propose to close this issue since the initial goal has been met, but
please open a new one to follow up on this.
Thanks!
Ludo’.
--- End Message ---