guix-patches
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[bug#60266] [PATCH] gnu: Add form.


From: Eric Bavier
Subject: [bug#60266] [PATCH] gnu: Add form.
Date: Mon, 09 Jan 2023 18:01:33 +0000

Hello Antero,

Thank you for the patch.  Sorry for the delayed reply, holidays and
such.  I was able to apply your patch and build a form package.

A few comments:

On Fri, 2022-12-23 at 01:15 +0000, Antero Mejr wrote:
> * gnu/packages/maths.scm (form): New variable.
> ---
> x86_64 only due to test failures on other platforms. Developers
> say other platforms are not "tier 1" supported:
> https://github.com/vermaseren/form/issues/426

This may be better as a comment near the `supported-systems` field,
along with a short summary of which tests fail on other systems.

> 
>  gnu/packages/maths.scm | 55 ++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 55 insertions(+)
> 
> diff --git a/gnu/packages/maths.scm b/gnu/packages/maths.scm
> index 050450e12c..08ddd2ecb4 100644
> --- a/gnu/packages/maths.scm
> +++ b/gnu/packages/maths.scm
> @@ -8161,3 +8161,58 @@ (define-public optizelle
>  provided for applications written in C++ and Python.  Parallel
>  computation is supported via MPI.")
>      (license license:bsd-2))))
> +
> +(define-public form
> +  (let ((commit "28e15eaf0856a0a012795298d6a4b570e764a8b1")

This commit is downstream from the 4.3.0 release, so we should include
in a comment the rationale for not using just the 4.3.0 release
tarball.

> +        (revision "0"))
> +    (package
> +      (name "form")
> +      (version (git-version "4.3.0" revision commit))
> +      (source (origin
> +                (method git-fetch)
> +                (uri (git-reference
> +                      (url "https://github.com/vermaseren/form";)
> +                      (commit commit)))
> +                (file-name (git-file-name name version))
> +                (sha256
> +                 (base32
> +                  "04i932lqwng2hmknvai1gmmb5j17rwrhlj11nr96w9bmj4sq736x"))
> +                (modules '((guix build utils)))
> +                (snippet #~(substitute* "check/examples.frm"
> +                             (("#pend_if valgrind\\?")
> +                              "#pend_if 0")))))

This snippet appears to be related to the specifics of this package
build?  E.g. if someone were to grab the source with `guix build -S
form`, they would not be able to have valgrind support, correct?  If
that's the case, perhaps it would be better to apply this substitution
in a build phase.

> +      (build-system gnu-build-system)
> +      (native-inputs (list autoconf automake doxygen ruby))
> +      (inputs (list bash openmpi))
> +      (arguments
> +       (list #:configure-flags #~(list "--enable-parform")

For some other maths packages that have both a serial and MPI versions,
we've usually provided two packages.  See e.g. `petsc` and `petsc-
openmpi`.  This can be useful if someone does not need a full MPI-
capable version of form, and the separation is easily achieved.

Also, I see the default compilation flags include `-march=native`,
which will most likely cause a problem with build farm substitutes or
`guix challenge`.  Could you figure out have to override these flags?
Guix usually assumes `SSE2` capabilities for x86_64 targets, iirc. 
Bonus points for enabling a "tunable" package (c.f. "Package
Transformation Options").

> +             #:phases #~(modify-phases %standard-phases
> +                          (add-after 'unpack 'fix-hardcoded-path
> +                            (lambda _
> +                              (substitute* "sources/extcmd.c"
> +                                (("/bin/sh")
> +                                 (string-append #$(this-package-input "bash")
> +                                                "/bin/sh")))))
> +                          (add-after 'build 'build-doxygen
> +                            (lambda _
> +                              (with-directory-excursion "doc/doxygen"
> +                                (invoke "make" "html"))))
> +                          (add-before 'check 'mpi-setup
> +                            #$%openmpi-setup)
> +                          (add-after 'install 'install-docs
> +                            (lambda _
> +                              (let ((doc (string-append #$output 
> "/share/doc/"
> +                                                        #$name "-" #$version
> +                                                        "/html")))
> +                                (mkdir-p doc)
> +                                (copy-recursively "doc/doxygen/html" 
> doc)))))))
> +      (home-page "https://www.nikhef.nl/~form/maindir/maindir.html";)

Maybe we should use https://www.nikhef.nl/~form/ instead? This is the
URL specified in the included manpage.

> +      (synopsis "Symbolic manipulation system for very big expressions")
> +      (description
> +       "FORM is a symbolic manipulation system.  It reads symbolic 
> expressions
> +from files and executes symbolic/algebraic transformations upon them.  The
> +answers are returned in a textual mathematical representation.  The size of 
> the
> +considered expressions in FORM is only limited by the available disk space 
> and
> +not by the available RAM.")
> +      (supported-systems '("x86_64-linux"))
> +      (license license:gpl3+))))

Thanks,
`~Eric

Attachment: signature.asc
Description: This is a digitally signed message part


reply via email to

[Prev in Thread] Current Thread [Next in Thread]