docutils does , as does html.
html element is::

  <a href="url">text</a>

there is url and there is text, it might be identical and it might not.

reST does::

  `text <url>_
  or text_

  .. _text: url

just because sometimes url and text are identical and might work for you  it is not a general solution.

as --input-encoding was already mentioned, what to do about --output-encoding, docutils
will have to honour that for text, but not for url, as url references something somewhere else.

am i wrong or simply lost ?



On 22 March 2014 18:25, Philipp A. <flying-sheep@web.de> wrote:
2014-03-22 15:56 GMT+01:00 engelbert gruber <engelbert.gruber@gmail.com>:

http://en.wikipedia.org/wiki/Uniform_resource_locator lists

A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
 a b c d e f g h i j k l m n o p q r s t u v w x y z
 0 1 2 3 4 5 6 7 8 9 - _ . ~

other characters must be encoded %20 for a blank, ...

at least that is what i understood.

that’s what the spec says, but everything except low-level code in browsers and servers accepts such urls when pasted into address bars, links with verbatim unicode in urls are clicked and so on.

i’d say that RST should support that as well.