On Wed, Sep 17, 2008 at 11:47 PM, Marco Antoniotti <marcoxa@...> wrote:
> On Sep 15, 2008, at 15:16 , Erik Huelsmann wrote:
>> On Sun, Sep 14, 2008 at 10:11 PM, Erik Huelsmann <ehuels@...> wrote:
>>> On Sun, Sep 14, 2008 at 9:23 PM, Marco Antoniotti <marcoxa@...>
>>>> You may want to have a look at (shameless plug)
>>> Interesting read. Actually, this is mostly how I solved it, except for
>>> the "platform specifier" part. I have an interesting use case however:
>>> ABCL allows pathnames to be nested: A pathname can point to an entry
>>> in a JAR file. In that case, the device - i believe - is a pathname
>>> itself, pointing to the JAR, the directory, name and extension all
>>> refer to a directory entry within the JAR.
>>> This could be a nice extention to your specification.
> Yes. This may be an interesting addition. But things would get quite
> hairy. Moreover, JAR files always sort of imply a "relative" naming scheme.
I've given this some thought, but if the jar file itself is considered
a device, then I don't see much of a difference between a path in a
jar file and a path in - say - Windows C:. There's one major
difference: the fact that you can make Windows C: paths the 'current'
path. However, from the point of a developer using Lisp, this fact can
be easily hidden: the *default-pathname-defaults* could easily point
into a jar file...
Am I missing a detail you *have* thought about?
>> An additional remark/question: have you considered making it a CDR
> Yes. But the code (and the HTML in particular) in in a state of flux now
> and I don't have the time to polish it. I am looking for helpers.
I haven't looked at the code. Is it built on top of existing path
functionality? Or is it a replacement to be incorporated by
implementors? Or is it a prototype implementation to be used for
testing? How much code is it? Would it be easy to complete? If it is,
how hard would it be to create a CDR for it?