From: Jason Mayer <slamboy@...>
Date: 2007-01-30T12:31:27+09:00
Subject: Re: Is my code using the Ruby way?

------=_Part_117440_19578574.1170127884902
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

On 1/29/07, ara.t.howard@noaa.gov <ara.t.howard@noaa.gov> wrote:
>
> hopefully you realize this code contains a race condition.  if you use a
> two
> step process to check that the database is created and, if not, to create
> it -
> you risk that another process might create it in between those two
> steps.  a
> cleaner way to do it is to use execptions:
>
>    harp:~ > cat a.rb
>    require 'sqlite'
>
>    class Database
>      SCHEMA = <<-SCHEMA
>        create table t(
>          id int,
>          data string
>        );
>      SCHEMA
>
>      def initialize path
>        @path = path
>        @db = SQLite::Database.new @path
>        setup
>      end
>
>      def setup
>        @db.execute SCHEMA rescue nil
>      ensure
>        validate_schema
>      end
>
>      def validate_schema
>        @db.execute 'select * from t limit 1'
>      end
>    end
>
>    Database.new 'db'
>
> here, the db is always created and initialized with the SCHEMA.  normally
> trying to create the table twice would throw an error, which is ingored.
> because ignoring this error might mask a real failure to set the db a
> simplistic method, in this case, is used to make sure the database looks
> like
> it's supposed to.  note that this whole thing is based on the knowledge
> that
> sqlite uses it's own locking to ensure atmoicity.
>
> in any case, the pattern of
>
>    "try it and recover if it blows up"
>
> is often a good solution where testing a condition and then acting on it
> cannot be done in one step to avoid a race condition.
>
> btw.  sqlite and ruby are a good combination - i use them heavily in my
> own
> work.


Just a general thank you for this detailed explanation - now I understand
better how to avoid race conditions (and I'm much more certain of what
exactly they are now).

------=_Part_117440_19578574.1170127884902--