Rails Caching in Agile Web Development w/ Rails4 - ruby-on-rails

Question

I'm working my way through the book Agile Web Development with Rails4, and I just read the (first) part about caching parts of a view to avoid overwhelming the database.
I've of course set the caching option to true in the config for the development environment.
The problem is that caching doesn't seem to be working.
Here is my app/views/store/index.html.erb file, exactly like the one given in the book, to enable caching :
<% if notice %>
<p id="notice"><%= notice %></p>
<% end %>
<h1>Your Pragmatic Catalog</h1>
<% cache ['store', Product.latest] do %>
<% #products.each do |product| %>
<% cache ['entry', product] do %>
<div class="entry">
<%= image_tag(product.image_url) %>
<h3><%= product.title %></h3>
<%= sanitize(product.description) %>
<div class="price_line">
<span class="price"><%= number_to_currency(product.price) %></span>
</div>
</div>
<% end %>
<% end %>
<% end %>
And here is the rails server log, clearly showing that the databse was accessed multiple times (although a line mentions caching) : http://pastebin.com/v2jGiHKL
Here is my app/views/store/index.html.erb file, where I tried something else for caching :
<% if notice %>
<p id="notice"><%= notice %></p>
<% end %>
<h1>Your Pragmatic Catalog</h1>
<% cache('caching') do %>
<% #products.each do |product| %>
<div class="entry">
<%= image_tag(product.image_url) %>
<h3><%= product.title %></h3>
<%= sanitize(product.description) %>
<div class="price_line">
<span class="price"><%= number_to_currency(product.price) %></span>
</div>
</div>
<% end %>
<% end %>
And here is the corresponding log, showing that caching was successful (as the database was not queried) : http://pastebin.com/ZTk9A9RA
Can someone explain why one seems to work and not the other, or how the first one should work ? Thank you :)
Note that in the book, it says that with caching enabled, reloading the store page shouldn't show new parts of the store/index.html.erb if the changes were made inside a cached block ; yet it does in both case. Any idea ?

Answer 1

The parameter you pass to the cache api is the caching key. The hashing function of that key is where the cached fragment will be stored, and where it will be looked for.
In your code (cache('caching')) the caching key is hard-coded, and never changes, so, unless the cache is invalidated (either manually in your code, or when its TTL has passed) - the same fragment will be sent to the client.
In the code from the book (cache ['store', Product.latest]) the key depends on the latest product. This means that if the latest product changes (another product was added, or updated) - the next call will automatically 'know' not to take the page from the cache, but to recreate the cache.
What is shown in the logs:
[1m[35mProduct Load (0.3ms)[0m SELECT "products".* FROM "products" ORDER BY "products"."updated_at" DESC LIMIT 1
Is actually the result of Product.latest, not of Product.all.
To sum up - both code fragments are cached, but in the example code, there is a (small) hit to the database to verify the validity of the cache, whether the page was cached or not.
If you are invalidating the cache yourself, you can keep the hard-coded version of the code, but if not, you better think of an invalidation scheme, whether it is like the example code suggests or something else.