Build system updates (call for testers!)



>>>>> "Rupert" == Rupert Swarbrick <rswarbrick at gmail.com> writes:

    Rupert> Raymond Toy <toy.raymond at gmail.com> writes:
    Rupert> automake; ./configure --with-something; make
    >> 
    >> I get message from automake about src/sharefiles.mk not existing.  but
    >> that seems ok.
    >> 
    >> Also, you removed the <lisp>-depends.mk files. Won't that mean that if
    >> I modify a lisp file, make won't rebuild maxima?

    Rupert> They should get automatically regenerated when Maxima gets built, I
    Rupert> think. When I test by deleting the one that got built, it reappears next
    Rupert> time make is run. They have a make dependency on maxima.system, so I
    Rupert> think they should even get regenerated at the right time. Does this not
    Rupert> work for you?

Ah, yes, the depends.mk file is regenerated.  So is sharefiles.mk.

    >> And, in fact, if I delete src/binary-cmucl, make no longer even tries
    >> to build the cmucl version.

    Rupert> That's worrying. For some reason, the debian cmucl package doesn't work
    Rupert> on my system, so I can't easily test this directly. But looking again at

You could grab a prebuild linux binary common-lisp.net.  (See
trac.common-lisp.net/cmucl for links).  They should work on a debian
system, but I haven't tested that.

    Rupert> Makefile.am, it seems that "all" should depend on $(FILES_TO_BUILD) (via
    Rupert> all-local), which should expand to $(BINARY_DIR)/maxima or
    Rupert> $(BINARY_DIR)/maxima.core depending on whether we have CMUCL_EXEC. Then
    Rupert> $(BINARY_DIR) should be binary-$(LISP_NAME), which (finally!) resolves
    Rupert> to binary-cmucl, meaning that "all" depends on binary-cmucl/maxima.core
    Rupert> or binary-cmucl/maxima.

    Rupert> Is there any chance you could track down where my chain of reasoning
    Rupert> breaks?

Yes, I'll look at it later.

Ray