[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: Some openldap fixes...

geef ff wat commentaar

You wrote:
> As the patch you provided is quite large, includes changes covering
> many issues, and is against a version not under active development
> (1.2 is only "maintained"), it is likely we need to discuss
> and determine an appropriate course of action.
I am aware of the fact that it's a 'dead' branch. However, when we started,
we could choose between 2.0-alpha, which was not quite stable at the time,
and the 1.2.x branch. Back then I had no idea that we would end up fixing
so much stuff.

However, that's why we tried to give an idea of the things we have fixed, so
other people could say "hey, that's cool for the next release" or "hey, I
could use some help with this or that". Although we
are both quite busy, we are willing to integrate stuff into the next
major release, because that's were most of our stuff would go. But keep 
in mind we're still working for a company, so it also needs to have
some value for the company.

Another 'problem' is that we didn't do it just for fun. We need the
features we've implemented. As major enhancements go into the developing
branch, we would be forced to use that branch. The code in a -devel
branch is not always stable enough for our use. In fact, even 2.0 crashes
after about 15 seconds of the load our own backends endure all day. I guess
I could look into this when I'm bored ;).

The next thing I want to say is that because all our fixes are
functional, we would not always agree on things. For example, 2.0
stores onelevel and subtree relationships in dn2id. The insertion
and removal of entries involves processing of ALL the parent nodes.
We have tried the same scheme with our new indices, but found that this
was too slow during changes, as well as during lookups. So we came up
with something faster. We place more emphasis on overall speed while you 
seem to be happy when it works.

And last but not least, we're doing this job next to our studies. If
we only had the job, we would have ported the changes to 2.0-alpha
a long time ago, just for the heck of it ;)


If at first you don't succeed, destroy all evidence that you tried.