The language issue you raise is in part solved by C# and Entity framework.

Because of the lazy way it works you can define a single select over a table, then apply extension methods, joins etc.. over the return of this select and you end up with a highly reusable clean easily tested data layer. Done right you get code like the following,

  var url = _urlRepo.GetUrl().OrderByCreateDate().ByCreatedUser("boyter");
  var url2 = _urlRepo.GetUrl().OrderByCreatedUser().ByDomain("geocities");

Or with joins,

  var users = _userRepo.GetUser();
  var locations = _locationRepo.GetLocation();

  var result =  from user in users
              join location in locations on user.locationid equals location.id && location.name = "Parramatta"
              select user;

I wrote about this a bit here http://www.boyter.org/2013/07/clean-repository-data-access-c... and it is to date the cleanest way I have come across in any language to deal with SQL. Compile time checked, reusable and testable.

Its the best ORM I have found so far across any language. I would dearly love to have it ported to other languages and frameworks.