On Jul 23, 5:21 pm, David Lee <[email protected]> wrote:
> Seeing that the @values instance variable is heavily used elsewhere in
> the code, we have to change the implementation to leave @values alone.
>
> Here is a proposal that will benefit in fixing the documented
> shortcomings (it will not benefit in detecting an internal
> modification of an attribute value using a ! method):
>
> class Model
>
> # allows user to see original values
> def saved_values
> @saved_values ||= {}
> end
>
> def []=(column, value)
> new_value = typecast_value(column, value)
> if @values[column] != new_value
> @saved_values[column] = @values[column] unless
> saved_values.include?(column)
> @values[column] = new_value
> end
> end
>
> def changed_columns
> saved_values.keys.find_all {|c| @values[c] != @saved_values[c]}
> end
>
> # allows user to see changes indexed by column name
> def changes
> ret = {}
> changed_columns.each {|c| ret[c] = [...@saved_values[c], @values[c]]}
> end
>
> end
This approach would allow you to see the original values for columns,
which may help some applications, but I can't recall any requests for
that functionality. I think changes similar to the ones you suggested
for Model#[]= so that it handles values after typecasting are
worthwhile, but I'm not sold on the general utility of the
saved_values feature.
I suggest that the saved_values feature be moved to plugin, which I'd
consider having as one of Sequel's standard plugins. Model#[]= should
probably be refactored to consider the value after typecasting. I'd
consider other refactoring to make the plugin easier to implement.
> With this approach, the only performance cost is in #changed_columns,
> and it's not that big of a cost.
> The only other downside is that the internal modification of a value
> problem is magnified here:
>
> # Problematic both before and after new implementation
> x = Model.new(:name => 'david')
> x.name.gsub! 'david', 'dave'
> x.save #=> no change
> x.name = 'dave'
> x.save #=> no change
>
> # Problematic only in new implementation
> x.name = x.name.gsub!('david', 'dave')
> x.save #=> no change, but would have changed in previous implementation
I wouldn't worry about these issues. If you modify a model instance's
values hash behind its back, all bets are off.
Jeremy
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"sequel-talk" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/sequel-talk?hl=en
-~----------~----~----~----~------~----~------~--~---
