[bug#58261] [PATCH 10/11] gnu: Add yaehmop-tightbind.

[Liliana Marie Prikler]

Am Freitag, dem 07.10.2022 um 15:12 +0000 schrieb David Elsing:
> Liliana Marie Prikler <liliana.prikler@ist.tugraz.at> writes:
> 
> > Am Montag, dem 03.10.2022 um 00:19 +0000 schrieb David Elsing:
> > > * gnu/packages/chemistry.scm (yaehmop-tightbind): New variable.
> > > ---
> > >  gnu/packages/chemistry.scm | 82
> > > ++++++++++++++++++++++++++++++++++++++
> > >  1 file changed, 82 insertions(+)
> > > 
> > > diff --git a/gnu/packages/chemistry.scm
> > > b/gnu/packages/chemistry.scm
> > > index cc26aa329a..a39deff4bf 100644
> > > --- a/gnu/packages/chemistry.scm
> > > +++ b/gnu/packages/chemistry.scm
> > > @@ -42,6 +42,7 @@ (define-module (gnu packages chemistry)
> > >    #:use-module (gnu packages documentation)
> > >    #:use-module (gnu packages flex)
> > >    #:use-module (gnu packages fontutils)
> > > +  #:use-module (gnu packages gcc)
> > >    #:use-module (gnu packages gl)
> > >    #:use-module (gnu packages graphviz)
> > >    #:use-module (gnu packages gv)
> > > @@ -817,3 +818,84 @@ (define-public coordgenlibs
> > >      (description "@code{coordgenlibs} contains algorithms to
> > > generate 2D
> > >  coordinates.  It has an emphasis on quality rather than speed.")
> > >      (license license:bsd-3)))
> > > +
> > > +(define-public yaehmop-tightbind
> > > +  (package
> > > +    (name "yaehmop-tightbind")
> > > +    (version "2022.09.1")
> > > +    (source (origin
> > > +              (method git-fetch)
> > > +              (uri (git-reference
> > > +                    (url
> > > "https://github.com/greglandrum/yaehmop";)
> > > +                    (commit (string-append "v" version))))
> > > +              (file-name (git-file-name name version))
> > > +              (sha256
> > > +               (base32
> > > +               
> > > "1x0d75m1hgdb411fiv7c5bwq1n4y0swrll0gigh8v5c73kjxrja0"))
> > > +              (modules '((guix build utils)))
> > > +              (snippet
> > > +               '(begin
> > > +                  (delete-file-recursively "docs")
> > This one looks suspicious.
> > > +                  (delete-file-recursively "examples")
> > As does this.
> > What gives?
> I did not want to check the documentation in detail and the examples
> contain some corrupted "compress" archives of program outputs. :)
> But now I build the documentation instead and remove the archives.
> There still is one problem: Even though I set FORCE_SOURCE_DATE=1,
> the
> output of pdflatex is still not reproducible. Do you have an idea why
> that could be? There are only 3 bytes different in the pdf.
> Alternatively, I could just not build the documentation as it is
> independent of the build scripts anyway.
pdflatex has troubles with reproducible builds, but I can't quite
recall what option makes it not so, but rest assured those are likely 3
very well known bytes.

> > > +                  (delete-file-recursively "viewkel")
> > > +                  (delete-file-recursively
> > > "tightbind/f2c_files")
> > > +                  (delete-file "tightbind/diag.c")
> > > +                  (delete-file "tightbind/cboris.c")
> > > +                  (delete-file "tightbind/abfns.c")
> > > +                  (delete-file "tightbind/lovlap.c")
> > > +                  #t))))
> > > +    (build-system cmake-build-system)
> > > +    (native-inputs (list gfortran))
> > > +    (inputs (list openblas))
> > > +    (arguments
> > > +     (list
> > > +      #:configure-flags
> > > +      #~(list
> > > +         "-DUSE_BLAS_LAPACK=ON"
> > > +         (string-append "-DPARM_FILE_LOC=" #$output
> > > +                        "/share/" #$name "-" #$version
> > > "/eht_parms.dat")
> > > +         "-DBIND_EXE_NAME=yaehmop-bind")
> > > +      #:phases
> > > +      #~(modify-phases %standard-phases
> > > +          (add-after 'unpack 'chdir
> > > +            (lambda _
> > > +              (chdir "tightbind")
> > > +              #t))
> > > +          (add-after 'chdir 'patch-fortran-functions
> > > +            (lambda _
> > > +              (substitute* '("mov.c" "prototypes.h")
> > > +                (("lovlap\\(") "lovlap_(")
> > > +                (("abfns\\(") "abfns_("))
> > > +              #t))
> > > +          (add-after 'chdir 'patch-cmake
> > > +            (lambda _
> > > +              (substitute* "CMakeLists.txt"
> > > +                (("project\\(yaehmop C\\)") "project(yaehmop C
> > > Fortran)")
> > > +                (("abfns.c") "fortran77/abfns.f")
> > > +                (("lovlap.c") "fortran77/lovlap.f")
> > > +                (("(set\\(PARM_FILE_LOC.*)\\)" all init)
> > > +                 (string-append init " CACHE STRING \"\")"))
> > > +                (("add_library\\(yaehmop_eht" lib)
> > > +                 (string-append lib " SHARED "))
> > > +                (("target_link_libraries\\(test_eht
> > > \\$\\{LAPACK_LIBRARIES\\}.*"
> > > +                  all)
> > > +                 (string-append all
> > > "\ntarget_link_libraries(yaehmop_eht "
> > > +                                "${LAPACK_LIBRARIES})\n")))
> > > +              #t))
> > > +          (add-after 'install 'install-eht-parms
> > > +            (lambda _
> > > +              (install-file "../tightbind/eht_parms.dat"
> > > +                            (string-append #$output "/share/"
> > > +                                           #$name "-"
> > > #$version))
> > > +              #t))
> > > +          (delete 'check)
> > > +          (add-after 'install-eht-parms 'check
> > > +            (lambda* (#:key tests? #:allow-other-keys)
> > > +              (when tests?
> > > +                (invoke "./test_eht"))
> > > +              #t)))))
> > > +    (home-page "https://github.com/greglandrum/yaehmop";)
> > > +    (synopsis "Yet Another extended Hueckel Molecular Orbital
> > > Package")
> > > +    (description "@code{YAeHMOP} contains a program and library
> > > for
> > > performing
> > > +extended Hückel calculations.")
> > > +    (license license:bsd-2)))
> > Looking at this it seems unclear whether you are building yaehmop
> > or
> > tightbind.  What's going on?
> I'm building the yaehmop library (with the source code in tightbind/)
> and the "bind" program (which I called yaehmop-bind) that uses the
> library.
> Is it better to just call the package "yaehmop" (which I now did)?
> The source also includes a separate visualization program "viewkel",
> which does not use the library and would belong in another package in
> my opinion.
Taking a cursory glance at the CMakeLists, calling it yaehmop is indeed
the correct thing.  I have no idea what viewkel is supposed to do tbh.