Andrew Sullivan wrote:
> On Wed, Aug 29, 2007 at 03:52:49PM +0530, Trinath Somanchi wrote:
>   
>> Hi ,
>>
>> Is this release an  Release candidate or an PORDUCTION Release .
>>     
>
> If it doesn't have RC in the name, it's a release.  We don't have
> non-production releases.
>
>   
We have fairly frequently been doing "pre-releases" with "pre" in the 
name, as opposed to having "RC" in the name.

That happened with both 1.2.10 and 1.2.11, where we knew there were some 
"sore spots," and it was actually fully expected that the "pre" versions 
would change before becoming production.

But it's fair to say three things:

- A release is a release; we don't let one out if it is known *not* to 
resolve some known bug. 

That's why there has been a fair amount of time between releases in the 
1.2 series, lately; there have been some bugs found that took some 
searching to properly "seek and destroy."

- Replication, as a distributed application, is highly susceptible to 
race conditions, and is an application of above-average complexity, with 
the result that sometimes bugs do linger for a time before being 
discovered.  If you look at the release notes in the 1.2 branch, you'll 
see that a whole lot of (generally small, often unlikely-to-bite) 
problems have been found and dealt with.

We obviously can't guarantee that there aren't still some bugs lurking 
out there.  However, there's a third thing to say...

- While we don't know that there isn't some lurking bug, 1.2.11 *does* 
have resolutions to all the issues found in earlier versions in the 
branch, and thus may be reasonably expected to be the "best" version 
that we have had, so far.
> That said, the project is young, and there is a history of nasty bugs
> in it.  Please test thoroughly before deploying.
>
> A
>
>   
I don't expect 1.2.11 to have introduced any new platform dependencies 
(caveat: except in the code that implements log shipping *testing*), but 
there weren't many reports in about how well it ran on a diverse set of 
platforms.  So if you're not running Debian (where I tested it), or 
whatever version of FreeBSD Jan is running, then the guarantees are a 
bit weaker than I'd like to offer. 

Certainly you should test it thoroughly before deploying it, as Andrew 
suggests.  That's particularly important if you're new to it; there are 
doubtless some "sharp edges" here and there that may 'bite' you, and I'm 
specifically not competent anymore to notice them ;-).  Practicing in a 
test environment can show you what to watch out for in production.

And if you notice any "sharp edges" that are not commented on in the 
documentation, let us know.  They may be:
a) Fixable problems, that don't need to be "sharp edges," or
b) Something that others ought to be aware of.