[Date Prev][Date Next]
RE: win 2000 problem?
> Question: Pierangelo, you stated that the "developer of back-sql is not
> maintaining it any more." Assuming openLDAP 2.1.16 passes our
> validation unit tests, we are committing to it heavily as our LDAP
> solution. Do you anticipate problems with back-sql support/fixes down
> the line, or is there another developer stepping up to support it? Any
> thoughts on that are appreciated as they have a bearing on our project.
> Question: Is "schemacheck off" the way to go for the time being? Is it a
> workaround for a bug, or?
> Having just worked through a 2.1.16 build of Back-SQL on Windows 2000,
> I'm willing to answer build questions for this rather specific
> configuration if I'm able.
I'll elaborate further on the rest of your message in another post.
Let me answer the direct questions first. When I took over back-sql
maintenance, the backend API was rapidly changing from 2.0 to 2.1,
so back-sql was about to be abandoned. I'm not using it in production
environment, and I've been playing with it for a while as an
experimental database. However I don't have the resources to maintain
it appropriately, in fact I've only the possibility to develop it
under Linux using IBM db2 or postgresql. Since many people are
complaining about problems with mysql, I might try, based on time
availability, to use this rdbms also. There's no way I can play
with it under w2k so platform specific issues should be dealt with
by willing users (all I can do is apply blind patches that do not
conflict with Unix build). In this sense, due to its intrinsic
experimental nature, back-sql might never become "stable", at least
because it requires a lot of effort on the meta-data side (and,
in my experience, on hacking of meta-data handling to achieve the
desired rdbms independence).
You may want to use "schemacheck off" with back-sql because it is
hard to ensure that meta-data will satisfy schema constraints now
that 2.1 is enforcing consistency much more than 2.0 was. Of course
I'd prefer to use "schemacheck on", but this puts very hard
requirements on meta-data design.
>From my standpoint, I'm going to try to preserve current functionality
for those that were using back-sql in 2.0; I'm not going, at the
moment, to dedicate too much of my spare time to it (read: forget
about new features unless they're quick and easy to add, or represent
a dramatic functionality/performance improvement).