symphonists / cachelite

when an extension does force a cache bypass

By enabling it, the cache does not get purged on entry save: they are
rather purged by a sperate script, which must be configured as a cron
job.

The root page (/) was always excluded from the cache.
This commit rewrites the whole logic and adds unit tests
in order to test that particularly hard to get right algorithm.

Fixes #38

Setting CURRENT_TIMESTAMP has a default value requires MySQL 5.6+.
I used conditional comments to make the default value active on 5.6
setups only

Fixes #37

There are not reason to fatally crash if the xsl can't be parsed.

Also look for item nodes with a id attribute

For the cache to be effective, the http cache header must be the same,
regardless of the cache status (hit or miss).

This commit remove references to HTTP/1.0.

This commit also makes sure that custom content type are properly sent

Fixes #27

This scopes out thing better.
Also, the cache does not need to be fetched before it's written to.

Fixed bug introduced in 1db4e33

Some extensions, like the member extension, could want some request to
never use the cache nor be cached.

Since it's not easy to tell which request should be bypassed, a delegate
is the best way to let other parts of the system decide.

Fixes #30

The old schema was not scalable since it only allowed for ~85 entries
and ~85 sections references per page (2 digits id + | = 3)

The new schema makes it so each row can contain a page reference, one
section id and one entry id. Multiple rows are created for all the
values, which makes it possible to save all ids from the page.

This commit also makes DB errors in this extension silently fail: there
is nothing worst than crashing a perfectly working page because the
cache settings could not be written.

Fixes #36

Fixes #31 does not respect the protocol

Multiple issue fixed on the method _in_excluded_pages().
It will now behave normally and respect : 
Excluding pages

Preferences. Each URL must sit on a separate line and wildcards () may be used at the end of URLs to match everything below that URL. /about-us/get-in-touch/ http://root.com/about-us/get-in-touch/ about-us/get-in-touch* /about-us/get*

This prevents a fatal crash of the extension. Let the site handle its
xml problem.

Version 1.3: Merge Integration