RE: API Programming questions

["Howard Chu" <hyc@highlandsun.com>]

> -----Original Message-----
> From: owner-openldap-software@OpenLDAP.org
> [mailto:owner-openldap-software@OpenLDAP.org]On Behalf Of Ugen

> >> That is not always possible. When the project is built on
> extensively
> >> modified server, rewriting it again is a hassle not worth doing.

> > Maybe extensively modifying the server was simply going the
> wrong way?
>
> No, it was the right and the only way. Working on custom
> product will do that to you sometimes:)

Having gone a long distance down this road with Symas' Connexitor product,
I'll have to side with Michael. The way to keep things maintainable is to
isolate your customizations to self-contained modules. Where interfaces for
your module are lacking in the main code base, you contribute patches back to
the main code base to create the necessary interfaces. By helping to improve
the public code base you make it easier to maintain your private code and
everyone wins. Keeping everything private only locks you into a dead end.

> > That's something misleading with Open Source: It sometimes looks
> > attractive to leave the code stream but it's not in the long run...

> > So your rant does not seems appropriate to me.
>
> And your sanctimony does not seem appropriate to me. The
> answer "upgrade to
> latest version" always struck me as mean, ridiculous and something
> worthy of Microsoft.

That comparison doesn't sit well with me. We turn around bug fixes here
faster than most commercial companies, and for the most part the latest
version is provably more stable than the previous one. (We also don't need to
halt all development for 6 monnths in a row to teach ourselves about secure
programming practices...)

> "latest version" only works when it is a drop in replacement
> for an old
> version. new
> OpenLDAP is NOT a drop in replacement in any shape or form. simply
> abandoning
> users of previous version is a bad customer service and a bad way to

The OpenLDAP project does not sell anything, so it has no customers, so what
customer service are you talking about? (My company [Symas] has many
customers, and everyone of them is thrilled with the service we provide them.
Hmm...)

> make people
> like your project. Good projects provide support and maintenance to
> older versions,
> especially something that was one less then current.

OpenLDAP 2.1 was released on June 9, 2002. Active support on OpenLDAP 2.0
continued through December 2002, with 3 or 4 maintenance releases, despite
the increasing divergence between the two code bases. In what way did anyone
get abandoned?

Take a look in the OpenLDAP bug tracking system - how many outstanding bugs
are filed against release 2.0? I count only 3. One (ITS#1174, aliases) was
against a feature that was *experimental* to begin with, and could not be
fixed without a major rewrite of the backend's search processing. If it was
important enough to someone to fix it sooner, then they should have stepped
up and done it. The 2nd (ITS#1590, startTLS response) is not so much a bug as
it is an ambiguity in the LDAP spec. The 3rd (ITS#1891, ber_flush failure)
was mostly fixed, with a final problem never reproduced and thus unresolved.

OpenLDAP is a community-supported project, driven by volunteers. Support
depends on all the members of the community to find bugs, submit bug reports,
and submit patches. If you're unhappy with the level of support the 2.0
release received, whose fault is that?

There are maybe a handful of regular contributors to this project, and every
one of them is busy with their life. How long do you expect anyone to
continue laboring over a broken system, before you finally realize it's time
to cut your losses and take a fresh approach? How do you justify your
expectation that any of them continue to spend precious time on *your*
behalf, keeping your old system working?

There's an old saying "the squeaky wheel gets the grease" which implies that
people who complain the loudest get the most help. Personally, I think the
people who just complain serve only to annoy the people trying their best to
do useful work. There's no porkbarrel here, if you want to make things move
you have to grease the wheel yourself.

  -- Howard Chu
  Chief Architect, Symas Corp.       Director, Highland Sun
  http://www.symas.com               http://highlandsun.com/hyc
  Symas: Premier OpenSource Development and Support