11.02.2014 22:20, Alexandre Benson Smith wrote:

> That's the point... we could have minor versions with short release cycle.

We did what could be done, see v2.1 and v2.5. They appeared exactly
because the codebase was not ready for v3-promised features and nobody
wanted to wait 5+ years between releases. It's no longer the case, however.

> So, if during this alpha stage the development team find some really
> serious problem with SMP on 3.0, we should wait for 4 ou 5 years until
> it's fixed ? Or do you think that a re-schedule (realease a 2.6 or
> anything you name it) with bug fixes, security fixes, new features, but
> using the old SMP architecture, is not an option ?

It is surely an option. However, with no serious problems re. SMP but
with huge efforts required for debugging/testing it, how could v2.6 be
possible? All the non-SMP fixes / features also require a lot of
testing/bugfixing. We either develop one thing or another, but not both.

> I think we live on a dynamic world that anything could and should be
> changed if it will be better.

As long as it's practically doable.

> For me will be much more important to free Firebird from the 27 chars
> restriction on object names, or a new wire protocol, than to have a new
> SMP architecture. If one say that he needs 5 years to implement a new
> SMP architecture, but 6 months to release a new version without the 27
> char limit you could bet that I would choose the 6 months version.

The limit is 31 chars in v2.5. I know many users would like to extend
it, although the Oracle example shows us that it's not really that
important as people tend to think.

> And, in the background the longer tasks could be developed without
> restricting the release of the new (not so major) versions.

In the background? Do we have a dozen of developers with free hands? ;-)

> As I said before, FB 2.1 (yes I don't even use 2.5 in production) fits
> me very well, but if I could put my hands on other minor features faster
> I would get some benefits from it.

I have nothing against the idea in general, I just want you to look at
the issue from another side. v3 is the next major release, it was
already delayed twice. v2.6 or whatever means delaying it once more, at
least for one year. If longer, you'll suggest v2.7... and some day I
will get tired of promising SMP to another part of users who is waiting
it since 2005.


Dmitry