Jess Robinson wrote:
 >> I'm created a producer_args key, server_version, that you can set to
3.23,
>> or 4.0, or any other version of MySQL. This influences the field type that 
>> we generate for boolean fields.
> 
> I couldn't quite figure out, what does this default to, if the user 
> doesn't supply a version at all?

That's the whole point. If the user (i.e. some random module that calls
SQLT but hasn't itself been updated) doesn't supply a version, then SQLT
should provide the existing behaviour not a new behaviour. Pedro stated
the opposite in the third sentence below.

>> Pre-4.1, MySQL producer will generate a tinyint. 4.1 and above we will 
>> generate a boolean. if no version is given, we generate boolean.
> 
> Any reason not to let the <4.1 output remain as it was? Then it's 
> somewhat backcompat.

> As for the general backcompat discussion: In general backcompat is nice, 
> when it comes to API/syntax/option args etc etc. I don't think it applies 
> to upgrading the SQL we output to cope with new versions of databases.

I don't believe the issue is to do with database versions. It's to do
with second- or third-party client code that hasn't been modified to
cope with the new SQLT release. For example, just because I rerun SQLT
to regenerate some of my application code doesn't mean I'm going to
rebuild a database running on some other server.

> We likely already have some features that won't work with some
> versions of databases that sqlt used to support.

Those would be what are called 'regression bugs' then :)  Especially if
you don't already know about them and have documented them.

Cheers, Dave

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
-- 
sqlfairy-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/sqlfairy-developers

Reply via email to