[horde] passwd H4 polishing for release - Requirements and directions?

Ralf Lang lang at b1-systems.de
Mon Jun 20 15:57:19 UTC 2011


Am Mittwoch, 15. Juni 2011, 13:17:16 schrieb Ralf Lang:
> Am Mittwoch, 15. Juni 2011, 12:47:43 schrieb Jan Schneider:
> > Zitat von Ralf Lang <lang at b1-systems.de>:
> > > Am Donnerstag, 9. Juni 2011, 20:23:00 schrieb Eric Jon Rostetter:
> > >> Quoting Ralf Lang <lang at b1-systems.de>:
> > >> > I've been toying around with the unreleased passwd from git.
> > >> 
> > >> Thanks.  Are you submitting back to git?
> > > 
> > > The changes are already in git.
> > > http://git.horde.org/horde-git/-/browse/passwd/?ws=1
> > > 
> > >> > For its main purpose, changing the user's horde account pw, passwd
> > >> > doesn't need to be a separate app (as said in the roadmap).
> > >> 
> > >> Yes, but I would hope we could keep it doing more than just that.
> > > 
> > > Me too.
> > > 
> > >> > But passwd is/was more generic and could be configured to change any
> > >> > password in a given infrastructure independent of Horde. Such a
> > >> > feature should be a separate app and not live somewhere under
> > >> > options or preferences. This could evolve into a generic
> > >> > password/credentials manager but is a completely different story.
> > >> 
> > >> And an interesting one, though I'm not sure who if anyone would want
> > >> to work on it (so it may never happen).
> > > 
> > > I had done some work out of tree for re-implementing eleusis from cvs.
> > > This has just stalled because there were more urgent tasks.
> > > 
> > >> > I am not sure what is needed for a proper passwd H4 3.2 release but
> > >> > I've found
> > >> > 
> > >> > the following items:
> > >> >  * convert the ldap backend to Horde_Ldap
> > >> >  * Make sure the composite backend uses the injector instead of the
> > >> >  base
> > >> > 
> > >> > class' factory method
> > >> > 
> > >> >  * remove that method
> > >> >  * make sure things like Horde_Vfs are included via injector
> > >> 
> > >> Would certainly be a good start (and work for me, since I use it with
> > >> the ldap backend).
> > > 
> > > We are currently using it with the hordeauth backend, and hordeauth is
> > > actually ldap. SQL has also been tested.
> > > 
> > >  * TODO: Add horde_test unit tests to passwd.
> > >  
> > >> > I think I could basically do that before H4.1 but passwd ships a lot
> > >> > of backends and I doubt I can build the infrastructure to test them
> > >> > all properly.
> > >> 
> > >> If you have horde, sql, and ldap complete, it could probably ship.  We
> > >> could see about removing some of the older, less used drivers too.
> > >> Others could be released if we at least think they should work, and we
> > >> can fix bugs with them as they are reported.  horde/sql/ldap should
> > >> cover most people...  Be nice if composite worked too, but probably
> > >> not a show stopper.  The kolab folks could probably help get the
> > >> kolab driver working.  I'd certainly not be opposed to dropping some
> > >> drivers (like pine, servuftp, etc) and shipping with untested ones
> > >> (like expect, procopen, vpopmail, vmailmgr, etc).  As long as we have
> > >> some well tested popular ones together (sql, ldap, etc).
> > > 
> > > [..]
> > > 
> > >> Some could just be dropped (until someone wants to pay for them?).
> > > 
[..] 
> > We should still convert the existing drivers to H4 blindly. Otherwise
> > the code would diverge too much over time. It's sufficient if the code
> > looks correct and parses. Users of those drivers can do tests with
> > production environments then.
> 
> [..]
> 
> > Jan.
> 
> Okay. I'll blindly convert the remaining drivers next.
> 
> Since I started the thread, I didn't touch any but the three "main" drivers
> discussed here, horde, sql and ldap. The latter uses Horde_Ldap now and
> will be tested.
> 
> If somebody with an ActiveDirectory setup volunteers for testing, I'll also
> address that use case.


Update:

Done:
 * Introduce an injector / factory for passwd drivers
 * New "horde" passwd driver using horde's configured auth backend
 * Convert the "sql" driver to Horde_Db (working)
 * Convert ldap driver to Horde_Ldap (working)
 * Removing of Passwd_Driver::factory
 * Converting PEAR_ERROR to Passwd_Exception

WIP:
 * Formal conversion of the remaining drivers 
   - I'd really like to see somebody trying out the Kolab Backend. It's 
basically LDAP but I won't convert it to Horde_Ldap or base it on 
Passwd_Driver_ldap without a test installation.
   - For most drivers I am sort of confident they might work.
   - The Pine backend looks like it won't work (Horde_Vfs)
   - vpopmail looks very similar to sql now - I think this should be merged or 
based on it if someone is really using this driver.

 * package.xml
   - I just created the file using the component tool and imported the most 
recent history. I did not edit any values. It guessed Chuck and Jan as leads 
while docs/CREDITS says it's Jan and Eric. Don't know where it pulls this and 
other information from

TBD (is this release critical?):
 * Add horde_test unit tests to passwd.
   - Need a closer look at the other apps what kind of tests they ship.
 * Trivial test of the composite driver 
  (any suggestions? Maybe two sql drivers with different DBs?)


What's left before we can try an alpha1 release? I think this might attract 
some users for bug testing and the alpha->beta->stable cycle should be much 
slower than with the mainstream horde apps (IIRC everything went from Alpha1 
to stable within roughly a month)



-- 
Ralf Lang
Linux Consultant / Developer

B1 Systems GmbH
Osterfeldstraße 7 / 85088 Vohburg / http://www.b1-systems.de
GF: Ralph Dehner / Unternehmenssitz: Vohburg / AG: Ingolstadt,HRB 3537


More information about the horde mailing list