RE: [Zope] ZopePacks: A Proposal for a new Zope Product...
-----Original Message----- From: Dan L. Pierson [mailto:dan@sol.control.com]
I agree as well. It might also be useful to include hooks in the pack format to support upgrading a product installation (e.g. converting all Squishdot postings to the new format :-)).
Currently objects can evolve themselves in their __setstate__ method. The pack operation doesn't really know much about objects and their classes, just records and transactions.
It might be useful to look at the docs for the RPM and see what kinds of things they were thinking about.
Absolutly, this is a great idea. Why think hard?
This is not to say that ZopePacks
should be anywhere near as comprehensive as RPMs. Rather, that they've thought of a number of issues that may or may not be germane the mechanisms in Zope.
If you do this you probably should also look at the Debian pkg stuff to see the features they found missing in RPMs. Again, you probably don't need all of it, but their analysis may be useful.
I can't immagine that either RH or Debian designed their packaging system without considering the idea of other people needing to use a similar tool. Perhaps there is a library that can manipulate either format that we can use that does all the hard work for us? -Michel
I can't immagine that either RH or Debian designed their packaging system without considering the idea of other people needing to use a similar tool. Perhaps there is a library that can manipulate either format that we can use that does all the hard work for us?
alien. :) elijah
Currently objects can evolve themselves in their __setstate__ method. The pack operation doesn't really know much about objects and their classes, just records and transactions. I'd like to point out that a __setstate__ method, like a __str__ method, cannot be created in a ZClass. Any particular reason why? How about providing the functionality of a super-"Renderable Base Z-Class" automatically to all ZClasses by creating default __str__, __setstate__, etc. methods that map to valid DTML method names? -- Stephen Pitts smpitts@midsouth.rr.com webmaster - http://www.mschess.org
participants (3)
-
elijah wright -
Michel Pelletier -
Stephen Pitts