Regenerate dependent views when migrations alter tables about activerecord-sqlserver-adapter HOT 8 CLOSED

When I added stronger view support to the adapter back in the day, I created a method chain to the standard #reset_column_information for ActiveRecord.

http://github.com/rails-sqlserver/activerecord-sqlserver-adapter/blob/master/lib/active_record/connection_adapters/sqlserver/core_ext/active_record.rb#L42
http://github.com/rails-sqlserver/activerecord-sqlserver-adapter/blob/master/lib/active_record/connection_adapters/sqlserver/schema_statements.rb#L324

This basically gives us our own column information cache before handing things back up to ActiveRecord. The benefits were that we included column information for views as well. Most schema statements that alter a table call remove_sqlserver_columns_cache_for(table_name). I'm sure that would be the hook point.

Lastly, I dont want to get into the A depends on T, depends on B situation. If a person's views structure is that complicated. Let them use :execute_procedure inside the migration and do the leg work on their own. I'm for a simple tested patch when you have the time :)

from activerecord-sqlserver-adapter.

Thanks. I'll take a look at those links.

The tree of dependencies is exactly the itch I need to scratch on our project. I'm trying to move us away from tools that handle that nicely (redgate), but do everything else worse than migrations (IMHO). Maybe I'll try and implement it as a module you can just drop into your project's lib or vendor directory. Could think of it as a sqlserver adapter plugin. Adapter plugins sound like a cool idea to me :)

from activerecord-sqlserver-adapter.

Me too, IE, method chain our method chain or a core method like remove_sqlserver_columns_cache_for. Very interested to see what you come up with! Cheers.

from activerecord-sqlserver-adapter.

Hi,

I'm running into a related issue where I've updated a table, but the Adapter has old columns from a previous migration in the list of attributes. Here's the table:
Column_name Type Computed Length Prec Scale Nullable TrimTrailingBlanks FixedLenNullInSource Collation
id int no 4 10 0 no (n/a) (n/a) NULL
client_id int no 4 10 0 no (n/a) (n/a) NULL
model_name varchar no 50 no no no SQL_Latin1_General_CP1_CI_AS
population decimal no 9 18 4 yes (n/a) (n/a) NULL
currency_code varchar no 3 yes no yes SQL_Latin1_General_CP1_CI_AS
first_landing int no 4 10 0 yes (n/a) (n/a) NULL
created_at datetime no 8 yes (n/a) (n/a) NULL
updated_at datetime no 8 yes (n/a) (n/a) NULL

and here's the attributes for the Model:
ree-1.8.7-2010.02 :001 > m = Model.new
=> #<Model id: nil, id: nil, client_id: nil, client_id: nil, model_name: nil, dimension_name: nil, dimension_value: nil, population: nil, currency_code: nil, custom_value: nil, created_at: nil, custom_value: nil, first_landing: 0, updated_at: nil, created_at: nil, updated_at: nil

How can I flush the column cache or reset column information? Thanks,

from activerecord-sqlserver-adapter.

That question is related to core ActiveRecord, it is a feature of it, not the adapter.

Model.reset_column_information

from activerecord-sqlserver-adapter.

Ah, thanks for the tip. I tried calling that method in the console, but unfortunately the stale column definitions are still being recognized by the database

ree-1.8.7-2010.02 :003 >  Model.reset_column_information

m = Model.new 
 => #<Model id: nil, id: nil, client_id: nil, client_id: nil, model_name: nil, dimension_name: nil, dimension_value: nil, population: nil, currency_code: nil, custom_value: nil, created_at: nil, first_landing: 0, created_at: nil, updated_at: nil, updated_at: nil>

I only posted here because I'm using both mysql and sqlserver adapters in different environment; sqlserver for dev and mysql for test. The test environment using the MySQL adapter shows the correct # of columns, while the SQL Server Adapter does not:

[mnyon@nyon ~/Documents/workspace/mroi (svn::39689:/apps) ] RAILS_ENV=test s-c
Loading test environment (Rails 2.3.5)
ree-1.8.7-2010.02 :001 > m = Model.new
 => #<Model id: nil, client_id: nil, model_name: nil, population: nil, currency_code: nil, first_landing: 0, created_at: nil, updated_at: nil> 

If there's another place I should submit this, please let me know. Thanks,

from activerecord-sqlserver-adapter.

I think you just have a simple problem of knowing what the code is doing. Migrations are run against a connection, not all connections & environments to different DBs. These queries are (1) outside the scope of this ticket and (2) outside the scope of the adapter, please ask questions for help on rails-talk, etc.

from activerecord-sqlserver-adapter.

Closing this old issue out. I think using sp_refresh is best left as (a) something the user knowingly needs to do when altering view dependent schemas or (b) part of another library that would like to take up the work in solving this problem in an abstract way for multiple adapters.

from activerecord-sqlserver-adapter.