On Thu, Jun 28, 2007, Jeff Johnson wrote:

> [...]
> If you want to be using sqlite3 instead of Berkeley DB, then please
> state directly.

Sorry, Jeff, at this time I still cannot state anything directly here
because I've still not even decided what my final targets are exactly.
I'm still in the "determine what is the best way to build and run RPM
5 in a cross-platform way" mode. As I'm just someone who can't ignore
half-working/half-broken solutions I usually have to jump in and either
nuke it or fix it -- even if later I will not even benefit from a fix
myself.

> There are much deeper engineering problems that need to be addressed
> if the goal is a high performing reliable sql database.

High-performance is currently not on my list. I'm happy if RPM is at
least "reasonable" fast with its database operations. But cross-platform
reliability is very high on my list. I do not care very much whether I
in the end have to run RPM with Berkeley-DB or SQLite as long as the
stuff works 100% reliable.

Although I've to admit that I'm a large SQLite fan and at last
since Thomas Lotterer recently failed horribly in creating a small
Berkeley-DB-based reliability test program (where not even the
Sleepycat/ORACLE guys were able to tell how to solve the problems),
I became even more sceptical about the reliability of Berkeley-DB.

Additionally, the KOLAB people also fight since years against the
stability of their Berkeley-DB based OpenLDAP database. All this
doesn't have to mean anything for RPM and its particular use of
Berkeley-DB itself, but I at least try to keep all possible doors open
for alternatives and having SQLite here is something I find sexy enough
to investigate further.

> endless "Don't panic ..." nit-picky messages

Sorry if I've bothered you with my "don't panic" commit messages.
No intended offending from my side, Jeff.

                                       Ralf S. Engelschall
                                       rse@engelschall.com
                                       www.engelschall.com