[asdf-devel] ASDF 2 issues

Juan Jose Garcia-Ripoll juanjose.garciaripoll at googlemail.com
Thu Apr 1 15:43:21 UTC 2010


On Wed, Mar 31, 2010 at 11:16 PM, Juan Jose Garcia-Ripoll <
juanjose.garciaripoll at googlemail.com> wrote:

> I have uploaded a newer version of the feature that allows two different
> syntaxes
>
> :logical-host "MY-HOST"
> :logical-pathname "MY-HOST:MY-BASE"
>
> The three pathnames that are defined are "xxx;SRC;*.*.*", "xxx;DATA;*.*.*"
>

I have just realized that the :logical-pathname case is indeed much more
complicated than what I had assumed: my implementation just set the given
translations for the given host, overwriting previous translations.

A real implementation of arbitrary logical pathname translations would
require a much more complicated code, probably based on something like james
anderson's utilities, requiring:
1) Analysis of the argument to :logical-pathname
2) Analysis and comparison of existing previous translations
3) Producing an intersection set, with the new paths
4) Sorting out conflicts.

But this indeed makes it more questionable the possibility of arbitrary
paths vs. a simple host controlled by ASDF.

Why? Because we need more complicated functions, and because these functions
have to be present in the code that is shipped with these translations, in
other words, having an implicit dependency on a functionality provided by
ASDF.

So I go back to the original proposal of :logical-host, which I insist is
just simple, verifiable, does not imply complicated logic and can be
exported to compiled code and distributed files.

Juanjo

-- 
Instituto de Física Fundamental, CSIC
c/ Serrano, 113b, Madrid 28006 (Spain)
http://tream.dreamhosters.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.common-lisp.net/pipermail/asdf-devel/attachments/20100401/ba425e3e/attachment.html>


More information about the asdf-devel mailing list