[Axiom-developer] Re: root chunks

[C Y]

--- address@hidden wrote:

> Besides the argument that the algebra files contain more than one
> root chunk there is another difficulty with choosing a connection 
> between filenames and root chunk names. At the present time I'm
> walking the system downcase all of the filenames. If there was a
> tight coupling between file and root chunk it would require that I
> also modify each file.

The only difficulty comes when extracting something other than the "*"
chunk - several files unique to that one chunk of code must be
generated by the tangle and compile routines.  I suppose I could have
the lisps generate completely random file names for these cases but I
would prefer to have the generated files relate back to the chunk they
came from for debugging purposes.  To illustrate the difficulty, let's
say I have two chunks:  "bootstrap1.clisp" and "startup"   Suppose I
need to tangle both of these out of the same pamphlet.  I know how to
specify them, but I must tangle them to a destination file for
compiling.  What name should I select?  In the case of bootstrap1.clisp
the obvious choice is the chunk name, but in the case of startup it's
not so clear.  Is it a lisp file, boot file, or what?  If I see that
file lying around in the directory, where did it come from?  On the
other hand, if I generate bootstrap1-startup.lisp, that works for
startup but makes a mess out of bootstrap1.clisp.  If I tangle them to
randomfoo1.lisp and randomfoo2.lisp that solves the problem but makes
tracing a given file back to its pamphlet origins a bit harder.

I would prefer to have the target files for the tangle process chosen
automatically rather than being user or developer specified - that's a
detail no one should have to care about.  The tangled source file is a
generated file.  I would like to have a sensible rule for doing so, if
one can be found.

> In general I feel that there should be no correspondence between 
> filenames and contents. The namespace and organization of files is
> not related to the namespace and organization of information.

In general, I agree.  In the specific case of intermediate generated
files, I would like the filenames to retain some relationship to their
origin to aid debugging.

Cheers,
CY


      
____________________________________________________________________________________
Fussy? Opinionated? Impossible to please? Perfect.  Join Yahoo!'s user panel 
and lay it on us. http://surveylink.yahoo.com/gmrs/yahoo_panel_invite.asp?a=7