[Zope-PTK] Workflow - security model

Steve Spicklemire steve@spvi.com
Tue, 5 Dec 2000 07:25:55 -0500 (EST)


Hi Jay,

>>>>> "Jay," == Jay, Dylan <djay@avaya.com> writes:

    Jay,> PS. On a mildly related note. We are not progressing with
    Jay,> Zope for our product despite the possibility of having to
    Jay,> reproduce some of Zopes functionality in Java (mainly
    Jay,> security and undo of some kind of database backend). The
    Jay,> main reason for this was lack of control over the code. We
    Jay,> have plenty of tools for doing software process using CVS
    Jay,> for doing code reviews and merging etc. I can see no
    Jay,> un-dodgy way to let code that resides in the the ODB to
    Jay,> participate in this. What is needed is a full file per
    Jay,> object based ODB implementation or something. This again
    Jay,> runs into the same problem of "What about bizzare pickled
    Jay,> object data?". And how would you match zope users to CVS
    Jay,> users? And versions? How would changes in the file system
    Jay,> get back into Zope?

This is the problem ZCVSMixin tries to solve. While it's certainly
not perfect at this point, it does make life a lot more bearable,
and I do know that lots of folks are using since I routinely get
feedback about it... and that only helps make it better. There are
two ways we use it:

1) Each developer works on a different system. They run their own Zope
and manage their own local CVS checkouts. It's just like normal CVS
development except your checking in and out Zope objects in the form
of xml and/or zexp files.

2) Each developer has a different (basically mirrored) 'area' in a
single Zope server. For each independent area there is a separate cvs
hierarchy that maintains their modifications/checkouts. This is 
of limited usefulness since it can't be done with ZClasses and
such... but it's useful for 'simple' content.

-steve