Hey!
i need to overwrite active record find to synchronize database access
You need to say super(*args). alias_method_chain is usually a more
prudent choice for this sort of stuff (and i assume you know that this
won't accomplish anything if you're using a pack of mongrels or
whatever - I suppose it would work if you had a single multithreaded
script, but I'm still not sure what the point would be)
Fred
Frederick Cheung wrote:
You need to say super(*args). alias_method_chain is usually a more prudent choice for this sort of stuff (and i assume you know that this won't accomplish anything if you're using a pack of mongrels or whatever - I suppose it would work if you had a single multithreaded script, but I'm still not sure what the point would be)
Fred
Problem is I'm using active_record without rails on a application that includes a mini http server (built on top of webrick) and a background task. Both operate on the same sqlite db. Because of concurrency issues with sql I'm trying to synchronize access to the database.
Iven with super(*args) i still get the error:
c:/ruby/lib/ruby/gems/1.8/gems/activerecord-2.0.2/lib/active_record/base.rb:1227:in `find_initial': undefined method `first' for nil:NilClass (NoMethodError) from c:/ruby/lib/ruby/gems/1.8/gems/activerecord-2.0.2/lib/active_record/base.rb:1514:in `send' from c:/ruby/lib/ruby/gems/1.8/gems/activerecord-2.0.2/lib/active_record/base.rb:1514:in `find_by_requesttype_and_requeststatus'
Frederick Cheung wrote:
You need to say super(*args). alias_method_chain is usually a more prudent choice for this sort of stuff (and i assume you know that
this won't accomplish anything if you're using a pack of mongrels or whatever - I suppose it would work if you had a single multithreaded script, but I'm still not sure what the point would be)Fred
Problem is I'm using active_record without rails on a application that includes a mini http server (built on top of webrick) and a background task. Both operate on the same sqlite db. Because of concurrency
issues with sql I'm trying to synchronize access to the database.Iven with super(*args) i still get the error:
Should have been obvious - it looks like you're not returning the
right value (ie the result) from find_by_sql. Also bear in mind that
find will call find_by_sql in some cases, so make sure whatever
locking mechanism you're using will support that. I've got a hint of a
doubt that by background task you mean a separate process, in which
case you're out of luck (because each process would have its own lock).
Fred
Frederick Cheung wrote:
Should have been obvious - it looks like you're not returning the right value (ie the result) from find_by_sql. Also bear in mind that find will call find_by_sql in some cases, so make sure whatever locking mechanism you're using will support that. I've got a hint of a doubt that by background task you mean a separate process, in which case you're out of luck (because each process would have its own lock).
I switched the locking mechanism to thread.exclusive, because a mutex doesn't work because find is calling find_by_sql. Added the returns.
def self.find(*args) begin #@@semaphore.synchronize { Thread.exclusive { #puts "find" return super(args)
} #} rescue puts "error: "+$! # @@semaphore.unlock end end
def self.find_by_sql(sql) begin #@@semaphore.synchronize { Thread.exclusive { #puts "find_by_sql" return super(sql) } #} rescue puts "error: "+$! # @@semaphore.unlock end end
By background task is a second thread, not a seperate process, so the locking mechanism should work.
Is every find call using find_by_sql to retrieve data? Would it be enough to syncronize just find_by_sql then?
thx
Frederick Cheung wrote:
Should have been obvious - it looks like you're not returning the right value (ie the result) from find_by_sql. Also bear in mind that find will call find_by_sql in some cases, so make sure whatever locking mechanism you're using will support that. I've got a hint
of a doubt that by background task you mean a separate process, in which case you're out of luck (because each process would have its own
lock).I switched the locking mechanism to thread.exclusive, because a mutex doesn't work because find is calling find_by_sql. Added the returns.
By background task is a second thread, not a seperate process, so the locking mechanism should work.
Is every find call using find_by_sql to retrieve data? Would it be enough to syncronize just find_by_sql then?
Almost all. a find with an :include doesn't in rails 2.0.2. This sort
of thing could change in the future though.
Fred
FWIW you can just use super [with no args] if it’s going to use the same arguments passed to the method. This works even if you modify the arguments like:
def label(object_name, method, text = nil, options = {}) text ||= method.to_s.titleize super end
RSL