Re: Thoughts on cx:zip

Florent Georges <fgeorges@fgeorges.org> writes:
>   If I am right, the actual content of entries can be retrieve either
> from external URIs (for instance the file system) or from the source
> port.  Is it possible to have the content for several entries from the
> source port?

Yes, source is a sequence and the base URI of each document is a
potential match for something in the manifest.

>   The orientation I took is to be able, in addition, to put content
> directly into the ZIP structure (what you call the manifest,) with an
> indication it has to be serialized as XML, HTML, text or binary.

Things that appear on the source port are serialized into the zip
file. That supports any of the serialization options, and should
probably also handle c:data elements that contain binary stuff.

>   For the function set, I have one function to read the ZIP structure
> (the manifest,) one to read a particular entry (actually 4 functions,
> to return either a document node, possibly from HTML, or a string or a
> base64Binary,) one to create a brand-new file and one to update an
> existing file.

That's covered by the existing cx:unzip step, I think.

>   Actually, the update function does not modify the file itself, but
> create a new one based on an existing ZIP file + some changes.  The
> idea is to not avoid side-effects.  But I know XProc is not afraid by
> side-effects.

Hmm. Yes, I suppose there's no harm in an option for putting the
resulting zip file in a different location.

                                        Be seeing you,
                                          norm

-- 
Norman Walsh <ndw@nwalsh.com> | Every vice you destroy has a
http://nwalsh.com/            | corresponding virtue, which perishes
                              | along with it.--Anatole France