| Message ID | YvJj5cXfFQs5RpR3@localhost |
|---|---|
| State | New |
| Headers | show |
Hello, I now pushed your updates to packages. Concerning the new additions, please try to add a description that is a bit longer than the synopsys and consists of complete sentences (which can be a challenge if the project homepage is too scarce, I can lend a hand). And it would be better to add the packages to different modules; many of them are added now with the aim of getting Sage, but are independent. So I would add primecount and gfan to algebra, palp to maths, python-lrcalc, python-primecountpy and python-pplpy to the same module as their "base packages", and python-memory-allocator probably to python-xyz. I would suggest to keep in sagemath.scm only the packages for which Sage has become upstream, such as flintqs. For tachyon, you use a patch and a phase; would only the latter not be enough? At worst by using substitute* on the Makefile? See the recent discussion on patches vs. phases. I did not yet have a look at the sagemath-... packages. And for pari-galdata, I wonder if there is not a better way so that it can also be used by the pari-gp package. The problem with pari-gp is that it cannot handle a search path with multiple directories, but that all data files need to be put into the same place. And all data files add up to about 200MB. One solution would be to add by default only the smaller data files galpol (10MB), nftables (8MB), galdata (52kB!) and nflistdata (4MNB). Actually, galdata is so small that we might just add it by default to pari-gp. It is not a perfect solution, but would improve the package at almost no cost, and apparently be useful for Sage. Andreas
Am Tue, Aug 09, 2022 at 03:40:53PM +0200 schrieb vicvbcun: > SageMath requires the FASL library? (I only know scheme) `maxima.fas'. > But it is only built when using ecl (See upstream here [0]). I don't > know if we need both sbcl and ecl. Arch uses both and puts the > lib/.../binary-* directories in separate packages[1], so maybe we could > use different outputs? Anyway, I have attached an rebased commit that > adds ecl. Okay, thanks! I do not use maxima (so it would be nice to get input from someone who does). To me it looks as if the different base lisps are more or less exchangeable for the goal of getting a running binary. So maybe it would be enough to build only the ecl version? The latest change was to switch from gcl to sbcl, in a commit by Guillaume, since maxima stopped building with the former; how about switching to ecl? Andreas
diff --git a/gnu/packages/maths.scm b/gnu/packages/maths.scm index e32a71100e..b250649a45 100644 --- a/gnu/packages/maths.scm +++ b/gnu/packages/maths.scm @@ -4357,6 +4357,7 @@ (define-public maxima (build-system gnu-build-system) (inputs `(("bash" ,bash-minimal) + ("ecl" ,ecl) ("gnuplot" ,gnuplot) ;for plots ("sbcl" ,sbcl) ("sed" ,sed) @@ -4365,7 +4366,8 @@ (define-public maxima (list texinfo perl python)) (arguments `(#:configure-flags - ,#~(list "--enable-sbcl" + ,#~(list "--enable-sbcl" "--enable-ecl" + (string-append "--with-ecl=" #$ecl "/bin/ecl") (string-append "--with-sbcl=" #$sbcl "/bin/sbcl") (string-append "--with-posix-shell=" #$bash-minimal "/bin/sh") (string-append "--with-wish=" #$tk "/bin/wish" @@ -4409,6 +4411,10 @@ (define-public maxima "--lisp=sbcl " "--batch-string=\"run_testsuite();\" " "| grep -q \"No unexpected errors found\"")))) + (add-after 'install 'install-fas + (lambda* (#:key outputs #:allow-other-keys) + (install-file "src/binary-ecl/maxima.fas" + (string-append (assoc-ref outputs "out") "/lib/maxima/" ,version "/binary-ecl")))) ;; Make sure the doc and emacs files are found in the ;; standard location. Also configure maxima to find gnuplot ;; without having it on the PATH.