| Message ID | 20221106090558.24507-1-mail@cbaines.net |
|---|---|
| State | New |
| Headers | show |
| Series | [bug#59073] gnu: r-minimal: Avoid referencing /gnu/store/[^-]+-glibc-[^-]+-static. | expand |
| Context | Check | Description |
|---|---|---|
| cbaines/comparison | success | View comparision |
| cbaines/git-branch | success | View Git branch |
| cbaines/applying patch | success | |
| cbaines/issue | success | View issue |
Hi, On dim., 06 nov. 2022 at 10:05, Christopher Baines <mail@cbaines.net> wrote: > * gnu/packages/statistics.scm (r-minimal)[arguments]: Split string involving > /gnu/store. It would imply rebuild many packages, no? master, staging or core-updates? > --- > gnu/packages/statistics.scm | 6 +++++- > 1 file changed, 5 insertions(+), 1 deletion(-) > > diff --git a/gnu/packages/statistics.scm b/gnu/packages/statistics.scm > index 2d4dbe4a31..43988ad00e 100644 > --- a/gnu/packages/statistics.scm > +++ b/gnu/packages/statistics.scm > @@ -470,7 +470,11 @@ (define-public r-minimal > "util-macros" > "graphite2")) > "|" > - "/gnu/store/[^-]+-glibc-[^-]+-static" > + ;; Be careful when including store paths in the > + ;; build script, since they might be treated as > + ;; references > + "/gnu/store" > + "/[^-]+-glibc-[^-]+-static" > ")/lib")) "")))))))))))) LGTM. Cheers, simon
zimoun <zimon.toutoune@gmail.com> writes: > Hi, > > On dim., 06 nov. 2022 at 10:05, Christopher Baines <mail@cbaines.net> wrote: > >> * gnu/packages/statistics.scm (r-minimal)[arguments]: Split string involving >> /gnu/store. > > It would imply rebuild many packages, no? master, staging or > core-updates? It does, my plan was/is to ask Ricardo if he could maybe include it when pushing the next big batch of r related changes? Chris
Hi Chris, On lun., 07 nov. 2022 at 10:32, Christopher Baines <mail@cbaines.net> wrote: > It does, my plan was/is to ask Ricardo if he could maybe include it when > pushing the next big batch of r related changes? I am preparing a Bioconductor upgrade, so it could be part of it. Ricardo, WDYT? Cheers, simon
Christopher Baines <mail@cbaines.net> skribis: > As this causes the following error when computing a cross-compilation > derivation for r-minimal and related packages. Which error? :-) Ludo’.
Ludovic Courtès <ludo@gnu.org> writes: > Christopher Baines <mail@cbaines.net> skribis: > >> As this causes the following error when computing a cross-compilation >> derivation for r-minimal and related packages. > > Which error? :-) Good spot, I've updated the patch/commit message. But it's this: → guix build --target=aarch64-linux-gnu r-minimal guix build: error: path ‘/gnu/store/[^-]+-glibc-[^-]+-static’ is not valid I'm not sure why this seems to only affect cross derivations. It's not the only problem that's happened for cross derivations though, so it might be good to expand the linting to cover them (maybe even for just one common target like aarch64-linux-gnu). Thanks, Chris
Hi, Christopher Baines <mail@cbaines.net> skribis: > → guix build --target=aarch64-linux-gnu r-minimal > guix build: error: path ‘/gnu/store/[^-]+-glibc-[^-]+-static’ is not valid Interesting bug! I pushed a fix and explanation: --8<---------------cut here---------------start------------->8--- commit 3bd4b92f55f40119349e39902a9b800de98040d2 build-system/gnu: Turn #:phases into a gexp when cross-compiling. Previously, we'd get this error: $ guix build --target=aarch64-linux-gnu r-minimal -d --no-grafts guix build: error: path ‘/gnu/store/[^-]+-glibc-[^-]+-static’ is not valid This is because the sexp would be passed as an input of the surrounding gexp in 'gnu-cross-build', and thus "/gnu/store/[^-]+-glibc-[^-]+-static" would be interpreted as a source file name, as in this example: scheme@(guix gexp)> #~(foo #$(list 'whatever "/gnu/store/[^-]+-glibc-[^-]+-static")) $11 = #<gexp (foo #<gexp-input (whatever "/gnu/store/[^-]+-glibc-[^-]+-static"):out>) 7f098badec30> scheme@(guix gexp)> (gexp-inputs $11) $12 = (#<gexp-input "/gnu/store/[^-]+-glibc-[^-]+-static":out>) Fixes <https://issues.guix.gnu.org/59073>. Reported by Christopher Baines <mail@cbaines.net>. * guix/build-system/gnu.scm (gnu-cross-build): When PHASES is a pair, pass it through 'sexp->gexp'. --8<---------------cut here---------------end--------------->8--- That said, this code is bogus: 1. We should refrain from using non-literal strings as patterns in ‘substitute*’. 2. We should call ‘%store-directory’ from (guix build utils) instead of hardcoding “/gnu/store”. Both of these should be fixed at some point. Thanks, Ludo’.
Hi, On Tue, 08 Nov 2022 at 14:09, Ludovic Courtès <ludo@gnu.org> wrote: > Interesting bug! > > I pushed a fix and explanation: Oh! :-) > That said, this code is bogus: > > 1. We should refrain from using non-literal strings as patterns in > ‘substitute*’. What does it mean “non-literal string” here? Cheers, simon
Hi, zimoun <zimon.toutoune@gmail.com> skribis: >> That said, this code is bogus: >> >> 1. We should refrain from using non-literal strings as patterns in >> ‘substitute*’. > > What does it mean “non-literal string” here? A string that is not a literal (or constant, if you prefer). This is OK: (substitute* f ((".*") …)) This is not: (substitute* f (((string-append "." (star))) …)) Ludo’.
Hi, On jeu., 10 nov. 2022 at 15:07, Ludovic Courtès <ludo@gnu.org> wrote: > A string that is not a literal (or constant, if you prefer). > > This is OK: > > (substitute* f > ((".*") …)) > > This is not: > > (substitute* f > (((string-append "." (star))) …)) Thanks for explaining. Cheers, simon
diff --git a/gnu/packages/statistics.scm b/gnu/packages/statistics.scm index 2d4dbe4a31..43988ad00e 100644 --- a/gnu/packages/statistics.scm +++ b/gnu/packages/statistics.scm @@ -470,7 +470,11 @@ (define-public r-minimal "util-macros" "graphite2")) "|" - "/gnu/store/[^-]+-glibc-[^-]+-static" + ;; Be careful when including store paths in the + ;; build script, since they might be treated as + ;; references + "/gnu/store" + "/[^-]+-glibc-[^-]+-static" ")/lib")) "")))))))))))) (define-public rmath-standalone