[Openmcl-devel] OSX install difficulties

[Francis Derive]

Dear all !

No difficulties - no more.

1) I use zsh - just new again and not much time to spare on this kind 
of thing, just let you know I had much help from
William Scott at www.chemistry.ucsc.edu/%7Ewgscott/xtal/zsh.html

So, I did "export" the CCL_DEFAULT_DIRECTORY 's value, and my % openmcl 
is OK. Thanks.

2) I definitely have to check how an archive expansion or a cvs 
download does populate the ccl directory issued from previous 
archive/cvs install.
     For example, just now as everything is fine at the end of it, I 
repeated the operation of a "tar xvzf" of the interface-database 
archive in a temporary folder which left the ccl directory unchanged ! 
- darwin-headers sitting with 188 KB on the disk. But "double-clicking" 
on the archive does expand in a "ccl.1" directory with a 5.8 MB 
darwin-headers on the disk !
This morning I couldn't expand with "tar xvzf" an archive downloaded by 
ftp in "text" mode ( Mac Os 10.3.5 ) as I had read it was the way to 
go, then downloaded in "Binary" mode and could "tar xvzf" the archive !
That's life here, and as I said I checked enough this afternoon - all 
the day long - to reach a very satisfactory situation.
About the discussion full archive versus (bin;source) I would say that 
if you do not want to compile the lisp image from source, the full 
archive is naturally strightforward.

3) I had to be cautious about the way to proceed with the "double" ccl 
directory - "current" vs "bleeding-edge". I had finished "building 
image from source" in both situation when I noticed there was an 
inconsistency about the date of creation of the "dppccl.image" files ( 
current vs bleeding).
Do you know what ? In a directory named "openmcl", I had a "ccl_current 
" - complete, with its final "dppccl.image" - to leave room to the 
"ccl" directory dedicated to the "bleeding-edge" installation. But my 
terminal was STILL pointing to the "ccl" directory even though its name 
in the Finder was now "ccl_current", and the "bleeding-edge" image had 
been "saved_application" in the "current" installation !
An openmcl heroe is something to be.
Now I have an "openmcl" directory, and an "openmcl-garage" where 
resides the version not in operation : in the operating "openmcl" 
directory there is only one "ccl" at a time.

  3) What else ? The doc. I had been confused with "Chapter 2.1.1 Source 
Code" example with the tar of a "bin" archive, while "Chapter 2.1.2 
Lisp image" example used a "source" archive. Now that I am fluent :) 
with "cross-dumping" I feel better - but when I think to the "Source 
code" I know that there is still no image - no dppccl.image, nor 
"ppc-boot.image" -, and when I think to the" Lisp image" I see the 
initial ddpccl.image which is in a "bin". So would you reverse the 
example ? Never mind.

4) So I went - easily, say half an hour for the dev version with a 
G4/450Mz/768MB - to the end of building a lisp image from the source, 
either with the current 0.14 or the 0.14-dev.
With 0.14 dev I went to " 2.5 Build the IDE" which I don't know yet to 
do about ..

4) I have a rest before going back and further read the doc which I 
need.

I hope to hear you soon again on this list, and I thank all of you 
three. Looks exquisite. Thanks !

Francis Derive
9 Allée de la Butte Rouge
91000 Evry France