[Top][All Lists]

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

[Axiom-developer] Re: accounts on tenkan

From: root
Subject: [Axiom-developer] Re: accounts on tenkan
Date: Mon, 11 Nov 2002 22:28:44 -0500

> Hello Tim and Tim,

address@hidden is my son. 

> I take it that you are not the same person but somehow
> related? <grin>
> Thanks for setting up the account. I was able to login
> and to check out files. I did
>   $ cvs ... co axiom
> and I am currently in middle of the process of downloading
> the whole set of project files. It is taking a rather long
> time to download - there are some BIG files in the zips
> directory. Here are a few initial questions. I'll probably
> have more when the download is done and I've had a change
> to look them over a little better.

Yes, there are some "serious big files" in zips at the moment.
There are 3 lisp platforms that Axiom is planned to run on, each
of which has advantages: CCL (Codemist Common Lisp), GCL (Gnu
Common Lisp) and CMUCL (Carnegie-Mellon University Common Lisp).
The current Makefile will build CCL from the lsp/ccl subtree and
GCL from the zips directory. The next version (or the one thereafter)
will also build CMUCL.

> I noticed that the directory structure is a little odd
>   axiom
>     new
>       new
>         license
>         lsp
>         src
>         zips
>       original
>         ...
> Did you intend to have two subdirectories names new?
> Is there a purpose for the outter most one?

The axiom/new/new is a mistake but there is no way to delete directories
thru cvs which is why i set up a practice site. The mistake was not
intentional (by definition, I guess).

The original directory was also a test case and can be ignored.

> BTW, the serious big files in zips seem to be mostly 
> Common Lisp stuff? How does that fit in compared to the
> Codemist Lisp?
> Finally, where should I start? Will the Makefile.pamphlet
> in the .../lsp/ subdirectory build a working CCL? It still
> looks a little "skinny" in terms of documentation... 

You should read the axiom/new/new/Makefile.dvi file (assuming xdvi
works on your machine. otherwise you might need to convert it to
pdf or ps. give me some feedback on formats). In theory you only
need change to axiom/new/new and type make. save a console trace
and we can fix what fails.

The Makefile is set up so you can type 
  make clean
and it will reset you back to the initial conditions. if you
make changes to any files in the source tree they will get
re-expanded when you next type 
i go to great lengths to cache work so the minimum work 
necessary to build a system occurs. However the make clean
will assume you want to start fresh and will destroy everything
(except changes in the src directory).

You can comment out the build of either ccl or gcl in the 
new/lsp/Makefile.pamphlet file (never change the raw Makefile as
it gets rewritten) and the change will happen in the next build.

Alternatively you can comment out the call in the new/Makefile.pamphlet
(which is the top level Makefile source and it treated as special)
but you need to recreate the Makefile by hand. Building the top level
Makefile will assume that you have the mnt/linux/bin/document command
and the mnt/linux/bin/tex/noweb.sty file. These are automatically
created by the 
command. Once the first make is done you can make changes in the
new/Makefile.pamphlet file and recreate it by typing:
  mnt/linux/bin/document Makefile
(note: not Makefile.pamphlet, just Makefile)

Except for the top level Makefile every other change to the system
should happen magically by typing make in the axiom/new/new directory.
If it doesn't let me know and I'll fix it. The Makefile.pamphlet tree
is supposed to document and automate all of the knowledge necessary
to build a system from scratch.

To test the result you can do:
  cd mnt/linux/bin
  echo "(+ 2 2)" | ./ccl
  echo "(+ 2 2)" | ./gcl

The correct answer is left up to the student but whatever it is
they should be = (but not eq).

> On my side: I've spent some of my spare time today trying
> to build a working version of TeXmacs under Cygwin. It is
> supposed to be posssible but still in "testing" mode
> according to their web site, but very few details were
> provided. So far no success. I've connected to their email
> list for further help.

As for TeXmacs just send a message to Joris van der Hoeven. He's
exceptional at responding (address@hidden). Andrey G. Grozin 
<address@hidden> is also quite responsive.

> Regards,
> Bill Page.


reply via email to

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