This was done because they needed to be split from one file to two, for efficiency, and it makes sense to have two files in their own directory. It probably makes sense to have each extension in its own directory anyway.
Tested both the inefficient backwards compatible interface and the present interface. Still more to be converted...
as is done with section anchors. This is hideous and ugly and evil, and I
do apologize, but it's the best we can do within XHTML's evilness.
This also corrects the breakage if you use really weird things like
< and > in your names, which would break the wikitext. It might be superior
to reject some such values entirely.
msg:cite_references_link_one and msg:cite_references_link_many, we already
use ��� (U+2190) and ��� (U+2192) in the main distribution so this shouldn't
cause any issues where this arrow doesn't exist in $popular_os_xyz
Text<ref>Foo</ref> <ref name=one>Bar</ref> <ref name=two>Zar</ref> <ref
name=two/>
format the "one" reference as if it were an anonymous reference, this is wise
because users frequnetly use named references for no practical reason and
displaying the list of backlinks intended for >1 items with only 1 item looks
ugly
'cite_references_link_', only direct references should be wrapped in <cite>
and since that probably won't apply to everything users put here I'm removing
it
* Documentation, marked functions as static where applicable
* Using ~ as a regexp delimiter instead of #, ��var like better..
* Killed PersistentObject
The quick<ref name=agility>NASA</ref> brown fox jumped<ref name=agility/> over the lazy<ref>Discovery channel</ref> dog
generate:
"The quick[1] brown fox jumped[1] over the lazy[2] dog."
inline and:
1. 1.1 1.2 NASA <!-- 1.1 & 1.2 are links -->
2. Discovery channel
in the references section, as opposed to:
The quick[1] brown fox jumped[2] over the lazy[3] dog.
inline and:
1. 1 2 NASA <!-- 1 & 2 are links -->
2. Discovery channel
like before
with the hook in Parser::clearState() makes sure the state of the Object is
kept in sync with the state of the parser, i.e. works correctly for dumpHTML
* give <ref> and <references> a css class, reference and references, respectively
* Code cleanup
* Lots of documentation, every function fully documented + code comments
* Split link() into linkRef() and linkReferences()
* Split key() into refKey() and referencesKey()
