Add :ignore_log_data option to #has_logidze

[DmitryTsepelev]

Fixes #97

[palkan]

Do we need to override this? If the column is ignored (in Rails 5), would Rails raise this exception for us?

I’d like to use as many Rails built-in functionality as possible

[palkan]

We shouldn’t call DB twice if we call log_data! twice.

Also, can we use pluck(:log_data) to avoid additional AR instance initialization?

[palkan]

Let’s move this patch into a separate file and add columns information manipulation (like in the original patch for Rails 5 or in ignorable).

[palkan]

Should we store the value in the attributes["log_data"]? As I understand, the following code reloads log_data (even if it present):

user = User.with_log_data.find(id)
user.log_data!

TBH, the API looks like a hack.

Maybe, it's better to add a method to reload log_data explicitly instead? Smth like:

def reload_log_data
  attributes["log_data"] = self.class.where(id: id).pluck(:log_data).first
end

This method could be a part of Logidze::Model.

Btw, we should not rely on id but on model's primary key instead:

self.class.where(self.class.primary_key => send(self.class.primary_key)).pluck(:log_data).first

[DmitryTsepelev]

Looks like id takes care about resolving primary key

[palkan]

It takes care of the value; does it take care of the key? (i.e. where(id: ...) still uses id)

Let's rename this method to reload_log_data, move to Logidze::Model and add a readme section. This method could also be useful to fetch the actual log_data after creating or updating a record.