[copying the rails list for more feedback] On 3/17/06, Jakob Skjerning <jakob@mentalized.net> wrote:> I believe you''re currently the man putting a lot of effort into the > Rails SQL Server adapter - is this correct?Somewhat. I am maintaining the SQLServer adapter but I''m not doing quite as much as I''d like with it and you certainly can''t call what I''m doing "a lot of effort" :)> I am in the process of persuading management to ditch our current setup > and get with Ruby on Rails. We currently have a 5GB SQL Server 2000 > database, which we''re pretty reluctant to change.My situation exactly.> So obviously management wants to know about the SQL Server connectivity > in Rails and I figure you''re the resident expert in that area. So if you > don''t mind, I''d like to ask a few questions. > > Do you use SQL Server on Rails production sites yourself?Yes. We run a health benefit analysis application with a modest number of concurrently active users and have been on Rails for almost six months now. We''ve experienced very few issues related to the adapter, although I''d love to see it get a bit more attention/refinement.> Do you know of others doing the same?I''m not positive but I believe most of the others using the adapter are using it for in-house apps. I could be wrong, though. Tom Ward would be a good person to ask or perhaps a quick ping to the Rails mailing list would be fruitful.> Is the SQL Server adapter up to par with the MySQL/PostgreSQL adapters? > (migrations, schema dumps, pagination, whatever).Yes and no, for the most part ;) It has the same capabilities (all of the things you mentioned) but is lacking some of the refinement present in the F/OSS adapters. This is mostly due to the sheer number of contributors on the F/OSS adapters and the relatively few on the proprietary ones but there are also some issues that stem from SQL Server itself. For instance, there''s no equivilent to the fairly common OFFSET modifier in other systems so paging requires a mess of ordered sub-selects combined with the TOP modifier.> What issues do the adapter currently suffer from?You can see a list of very specific, mostly minor issues here: http://tinyurl.com/nxkve From a more general level, the adapter needs someone to love it and dedicate a few weeks of time to really refining it. There''s plenty of room for making the adapter better.> Has there been made any performance testing/comparisons of the adapters?Not that I''m aware of.> Is there any particular linux distribution you would recommend for the > webserver, when the SQL Server is version 2000 running on Windows 2000.We run our production and staging environments on FreeBSD and have multiple development machines running Fedora Core 4. I''ve shyed away from Debian based distros because I''ve heard there are multiple issues with the Ruby packaging. I''ve experienced very few issues on the *NIX side of the adapter once working FreeTDS/unixODBC/RubyDBI pieces are in place. Pretty much every distro has had these packaged (and well) for quite some time now. FreeBSD''s packaging of these components is exceptional.> Thanks in advance for your time :)No prob. I wish the situation was more rosey. As crazy as this sounds, I''m planning on pushing on Microsoft to either pony up a part-time developer or put up some money to get someone on it. I''m 99% sure they''ll decline but it''s worth a shot and I don''t see a good argument for why customers who purchase SQL Server should not be able to have excellent support for it in today''s leading web environments :) Ryan Tomayko rtomayko@gmail.com http://naeblis.cx/rtomayko/ http://lesscode.org/
On 3/17/06, Ryan Tomayko <rtomayko@gmail.com> wrote:> > Do you use SQL Server on Rails production sites yourself?> > Do you know of others doing the same? > > I''m not positive but I believe most of the others using the adapter > are using it for in-house apps. I could be wrong, though. Tom Ward > would be a good person to ask or perhaps a quick ping to the Rails > mailing list would be fruitful.We use SQL Server on rails production sites, and have been doing so for almost a year (though we started with our own custom adapter rather than the official one). About 9 months ago we moved over to the official adapter, and helped Ryan implement functionality missing within it (though I''ve been pretty selfish in this, mainly sticking to problems that actually affect me). Since then we''ve had no real issues with the adapter.> > Is the SQL Server adapter up to par with the MySQL/PostgreSQL adapters? > > (migrations, schema dumps, pagination, whatever). > > Yes and no, for the most part ;) It has the same capabilities (all of > the things you mentioned) but is lacking some of the refinement > present in the F/OSS adapters. This is mostly due to the sheer number > of contributors on the F/OSS adapters and the relatively few on the > proprietary ones but there are also some issues that stem from SQL > Server itself. For instance, there''s no equivilent to the fairly > common OFFSET modifier in other systems so paging requires a mess of > ordered sub-selects combined with the TOP modifier.Can''t add anything that Ryan hasn''t touched on. Most of the functionality in ActiveRecord seems to be driven by people working against MySQL or similar. We then strive to port this over to SQL Server, but in some specific cases (particularly pagination) the solutions we have to use aren''t exactly ideal. The good news is the suite of unit tests in ActiveRecord is extensive and continually growing. There''s significant scope to refactor and improve the adapter code with the confidence we''re not breaking things. Some of the issues derive assumptions made in active record, so might involve changes that run deeper than just the actual adapter.> No prob. I wish the situation was more rosey.It''s not that bad. It works, people use it, the code is available (and not too difficult) to improve, and we''ve got a good test suite to prove we''re not making things worse.> As crazy as this sounds, > I''m planning on pushing on Microsoft to either pony up a part-time > developer or put up some money to get someone on it. I''m 99% sure > they''ll decline but it''s worth a shot and I don''t see a good argument > for why customers who purchase SQL Server should not be able to have > excellent support for it in today''s leading web environments :)That does sound crazy ;) Tom