[ZODB-Dev] packaging zodb in eggs.

[Jim Fulton]

Chris McDonough wrote:
> I'm taking a stab at packaging the various pieces of ZODB as eggs.   One 
> of the things I'd like to do is to separate the packages of ZODB  that 
> are currently shipped together (BTrees, ZODB/ZEO, persistent   
> transaction, ZConfig) into separate packages, because each may  arguably 
> be useful outside of what we call "ZODB" now.
> 
> One of the issues I've run into is that there are compile-time  
> dependencies between packages mainly due to C header files. For  
> example, the BTrees package assumes that it will be able to find the  
> cPersistence.h file, which currently ships with the "persistent"  
> package.  I wonder if I should bother breaking the persistent and  
> BTrees packages apart as separate eggs.  I suspect not.
> 
> I realize BTrees aren't terribly useful outside of persistent  storage, 
> so reasonable response would be to say "don't split pieces  of ZODB up 
> into separate subpackages unless they don't share much/any  code".  In 
> that case, I might create a "persistence" egg that  includes both the 
> 'persistent' and 'BTrees' packages.  The "ZODB" egg  might include both 
> ZODB and ZEO (these are linked inextricably) which  depends on the 
> "persistence" distribution.  Transaction should be  dependency-free (but 
> probably isn't, I haven't gotten that far yet).
> 
> So in any case, for ZODB, if I punted on the "one-package-one-egg"  
> relationship, we'd wind up with (for example):
> 
> persistence-3.6.0.egg (which would include persistent and BTrees  modules).
> zodblib-3.6.0.egg (which would include ZODB and ZODB packages).
> transaction-3.6.0.egg (which would include only the transaction  package).
> zconfig-2.3.1.egg (which would include only the ZConfig package).
> zodb-3.6.0.egg (which might be an empty "dependency" package that  
> depended on persistence, transaction, zodblib, and zconfig).
> 
> Anyone care?

Yup.  However, I think before we can package these separately, we need to
fix the dependency graph. in particilar we have a lot of cycles, which is
bad.  For example. ZODB depends on ZEO. persistent and transaction depend on
ZODB.  I think that this needs to be cleaned up first.  This is not a small
job.

BTW, the turbogears folks are interested in using transaction.  The transaction
package has shallow dependencies on ZODB.  A nice start would be to release
a separate transaction egg that doesn't depend on ZODB.  (Hint, we'll need a separate
weak-set egg that transaction and ZODB depend on and we'll need to decouple the
transaction exceptions from ZODB.)

Jim


--
Jim Fulton           mailto:jim at zope.com       Python Powered!
CTO                  (540) 361-1714            http://www.python.org
Zope Corporation     http://www.zope.com       http://www.zope.org