rails-sqlserver (Rails SQL Server)

This is handled here:
http://github.com/rails-sqlserver/2000-2005-adapter/commit/ad83df821022a7ff5297ba0a1ceee5335093a7a7

Wooops, wrong URL... the concept for this request is applied here:
http://github.com/rails-sqlserver/2000-2005-adapter/commit/ad83df821022a7ff5297ba0a1ceee5335093a7a7

Your concept is applied here:
http://github.com/rails-sqlserver/2000-2005-adapter/commit/4d208acc503bfb31f67857dfc1acae48e857d954

Going thru some forked commits and found this. Question: Does this solve a particular issue? When I created these, I made them class for the connection pools. So if a Apache/Passenger instance/fork was launched and it had a connection pool of 10, then expensive schema reflection happend once for that particular app instance/process. Would love more feedback on if this is something the core adapter should do.

Your concept is applied here:
http://github.com/rails-sqlserver/2000-2005-adapter/commit/4d208acc503bfb31f67857dfc1acae48e857d954

Your concept is applied here:
http://github.com/rails-sqlserver/2000-2005-adapter/commit/4d208acc503bfb31f67857dfc1acae48e857d954

There are a few things wrong with this. First the class attribute access "enable_default_unicode_types" is for migration support so that you can just say :string and a national/unicode nvarchar(255) is made. Using that as a way for getting quoted_string_prefix() is problematic. For instance, this accessor is false by default in my app and yet I use unicode columns.

Second, you have added raw unicode data. I had it that way in the adapter a long time ago, but had to change it to support ruby 1.9. This change in the test cases would break under 1.9.

I am having trouble understanding this one. Can you detail what this is solving and/or describe a test case?

This is "real" interesting. I would like to know more about it on the Google list if you could. Other testers including myself on 2000/2005/2008 with unixODBC have not see this issue? Are you compiling with +utf8?

Version 2.2.18 and forward have support for 2008 too. At some point too we will have better limit/offset code for 2005/2008 too.

FYI, we are both on gemcutter and rubyforge under the proper canonical name of "activerecord-slqserver-adapter". Obviously, gemcutter is now the preferred gem site given all the recent news with Github and RubyForge.

This will make it in the next release. I wrote some regression tests too. http://github.com/rails-sqlserver/2000-2005-adapter/commit/1a619a6ccd25127bdb8c5522651213bc97a6a038