Re: Need for 0.83 as stable version

Sebastien FLAESCH wrote:
> We experienced problems with 0.82 that are fixed in 0.83, but
> our customers hesitate to use a "non-stable" version.

I could be persuaded to release what we have as 0.83.  I have ambitions
for db-lib, but I agree that much has changed since January 2008.  

If you would like to speed up the release of 0.83, here's a list of things
that need to be done:

	*  update NEWS
	*  review UG, update 

I think it's helpful to quantify the changes in terms of lines of code and
API calls affected.  

> Maybe some roadmap / plans on the website would help, just to
> give an idea of the future versions (availability, features).

The roadmap is, "as needed"; the schedule is, "time permitting".  

Anyone who wants more certainty is welcome to lend a hand.  That doesn't
necessarily mean coding, although code's a plus.  Detailed bug reports
save me investigation time, and I don't know of one that went unanswered. 


Frediano said:

> The problem is that usually we release too stable releases :)

I disagree.  I know, "release early and often" is the mantra, and surely
Linus knows more than I do about building a community and engendering
excitement amongst developers.  That said, a database communications
library is not something many people look forward to upgrading.  It's a
critical piece of infrastructure, linked to by still more critical pieces.
 Upgrading may involve rebuilding Perl, sqsh, PHP, any number of C or C++
programs, and no small amount of testing.  

Consider how many ML messaages refer to 0,64 even today, code that was
superseded 18 months ago!  If we'd had 6 releases instead of 1 in the
meantime, that would only lead to more confusion and segmentation in the
user community.  

Regards, 

--jkl