This changes the previously static lastsecid into an instance variable,
this leads to consistent section edit ids when the xhtml renderer is
used more than once in a request. This makes it possible for test cases
to know the exact section edit id for an instruction array.

can someone with better regex skills please revise this?

This reverts commit 1ca2719c as it is
now superseeded by a6b82e436e3d68a42a6556165d6aaf9249db44cd

This fixes a problem where JavaScript could be introduced through
specially crafted RSS feeds on a lower level than the commit from
yesterday (1ca2719c)

This also fixes a problem where JavaScript links could be introduced by
specifying it as an RSS URL: the resulting error message displays a
link to the broken feed URL. This patch makes sure there's no working
link for unknown protocols.

This fixes a security vulnerability where an attacker could introduce
JavaScript links into wiki pages by including a prepared RSS feed.

This changes the cache logic for metadata. It introduces a new mode that
tries to avoid rendering the page again for simple requests but still
updates the metadata when the page has been changed (but not when the
cache timeout has been reached or purge is used). It simply compares the
time of the last rendering with the last modified time of the page.

The old boolean $render parameter has been changed into an int with
three possible values. Compatibility for the old parameter is provided
using a check with is_numeric using the following mapping:
- false is still don't render (0 is the new value for that)
- true is using that new render logic which means that many plugins will
  still work unchanged even if they request a lot of data using
  $render=true (1 is the new value for that providing full compatibility
  in the case 1 has been used instead of true)

The default value for p_get_first_heading is now that new simple cache
logic, the default value for getting metadata is the cache logic which
should be used with care but is the only way to request (rendered)
metadata that can change because of plugin installations or upgrades.

Anika's merge did not pul in the individual patches as one would expect.

Then I messed up when trying to fix this by merging with danny's repo
again but used the wrong branch. So we're still missing two patches. To
have them apply cleanly I have to revert Anika's merge here. Another
merge for the missing two patches will follow.

This reverts commit b17e20ac.

Like others, there's no need to define $sectionedits as private, and it prevents plugins that ought to handle headers from working and require a lot of ugly workarounds (such as header2).

If there's a concern, "protected" may be another choice.

As of VIM 7.3 it is no longer possible to specify the encoding in the
modeline. This gives an error message whenever such a file is opened,
thus this commit removes the enc setting from the modeline.