[Zope] ZCVSFolder + plain text export format

[Marius Gedminas]

Hi,

We use ZCVSFolder here at www.CodeWorks.lt extensively, and we've been
contributing patches to it.  One of the main chores with ZCVSFolder is
the unreadability of XML-exported files (random object IDs, random
reorderings of document nodes, escape sequences in document bodies
etc.).  This makes the XMLs basically unreadable, undiffable and
unmergeable.

As an alternative we developed an alternative export format, which
resembles RFC-822 a lot.  Here's an example of an exported Python
script:

  Bindings: container='container'
          context='context'
          namespace=''
          script='script'
          subpath='traverse_subpath'
  Id: language_buttons
  Local-Roles: mg=('Owner',)
  Meta-Type: Script (Python)
  Owner: acl_users
          mg
  Parameters:
  Permissions:
  Proxy-Roles:
  Title:

  # Example code:

  # Import a standard function, and get the HTML request and response objects.
  from Products.PythonScripts.standard import html_quote
  request = container.REQUEST
  RESPONSE =  request.RESPONSE

  # Return a string identifying this script.
  print "This is the", script.meta_type, '"%s"' % script.getId(),
  if script.title:
      print "(%s)" % html_quote(script.title),
  print "out", container.absolute_url()
  return printed

Advantages:
- This format is compact, easily readable, diffable and mergeable
- It contains only relevant object attributes, instead of being a dump
  of internal representation (e.g. storing both source code and
  base64-encoded compiled bytecode for Python scripts in exported XML
  files)

Disadvantages:
- Export of different object types is no longer automatic, you have to
  write code for every supported meta-type (only about 20 lines of
  Python for every supported meta-type on average).
- The code may drift, e.g., when new attributes are added into Zope
  objects but import/export is not updated.  This may cause
  maintainability problems.

I am now starting to think about future possibilities.  If such a
representation would get officially endorsed and included into Zope
proper, and import/export routines were added to the objects themselves,
it would eliminate the second disadvantage (code drift).  There would
also be a standard interface to get to a plaintext representation of
both data and metadata, more extensive than document_src().

The possibilities are endless.  It would be nice to be able to connect
to a running zope instance via FTP/whatever and get and edit objects in
this format.  This would allow the users to see/change metadata also.
There would no longer be a need for PUT_factory external methods for
creating new objects of various kinds.

What do you think about this?

BTW if anyone wants to see the actual code, take a look into ZCVSFolder
CVS repository (cvs.zcvsfolder.sourceforge.net module zcvsfolder),
mg-patches branch.  It will be merged into main branch in a week if
Steve Spicklemire approves.  I suppose I could make the code available
on the web somewhere if there's any interest.

Marius Gedminas
-- 
If C gives you enough rope to hang yourself, C++ gives you enough rope
to bind and gag your neighborhood, rig the sails on a small ship, and
still have enough rope left over to hang yourself from the yardarm.