plinehan wrote:

> ---------------------
> 2. Also quite important.
>
> Move the code base to 1 (preferably the gcc)
> compiler.

Not a good idea. Using several different compilers
uncovers problems, keeps the code cleaner and
avoids an unnecessary dependency.

> Having an Open Source project
> that relies on a Microsoft compiler does little
> for "street cred".

We don't depend on the Microsoft compiler. We
use it on Windows. We use gcc on Linux and Mac,
the Forte compiler or gcc on solaris, etc.

> It might also help lead to
> 2.5 which is support for the BSD's <insert
> your favourite OS here>.

We do support the various BSDs. In fact the problems
we have had with portability came from overuse of the
gcc stdlib which turned out to have horrible problems
with upward, downward, and sideways compatibility.

>
> I do also appreciate that a significant effort
> is underway to unify the code base, and have no
> more "Classic" or "Superserver", just something
> that will work anywhere. That should extend to
> the compiler being used.

Ideally we should work with anybody's compiler.

> 3. Not sure here. Would a write-ahead log
> give (some) users more confidence? Would it allow
> for PITR?

Add something unnecessary that increases the I/O load
to correct a possible problem of perception? Hmmmm....
Nbak allows PITR.


Regards,


Ann