Rob Anderton

Given that this patch fixes what is really a bug in ActiveRecord, doesn’t break existing code and is now the third such patch to address the issue of constructo...

My IPAddress example is probably to blame. The differences are:

• :constructor is called when instantiating the aggregate object and is therefore passed all of ...

Attached is a patch for composed_of aggregations that does the following:

1. It adds a new :constructor option that takes a symbol or a proc that will be used to...

I agree it's not a caching issue - it is a class reloading issue and as I showed in my last post it goes back at least as far as Rails 1.2.5.

To quote a Giles ...

We seem to be moving onto a different issue and one I've always assumed was expected behaviour. I'd thought that by reloading a class' definition you were effec...

Given that both the memcache and file stores (thanks to Ryan's other patch) will marshal objects, why is it such a horrible idea to use marshalling on the memor...

The Rails::Configuration class initialize method does not call the default_cache_store method to assign a default value to the cache_store attribute.

If no ca...

Updated patch attached: this doesn't touch the primary keys at all, it just adds an :unsigned option to migrations.

I have some other, non-MySQL specific, ide...

This was originally discussed but not resolved by others on the old Rails trac:

http://dev.rubyonrails.org/ticke...

I've attached a patch (with tests) that d...