Re: character entities in UTF-8 files

From: Gregg Reynolds (
Date: Wed Jul 13 2005 - 10:37:22 CDT

  • Next message: Peter Constable: "RE: Regarding Correct Display of Extended Latin Devanagari"

    Peter Kirk wrote:
    > On 13/07/2005 00:52, Gregg Reynolds wrote:
    >> ... From the Unicode perspective, a sequence of characters like
    >> é is just a sequence of 5 distinct characters with no further
    >> semantics. Interpreted in accordance with XML, however, such a
    >> sequence *must* (not "may") be interpreted as e acute. Note that (if
    >> I'm not mistaken) such interpretation logically precedes other
    >> parsing. That is, an XML parser will first interpret (i.e.
    >> substitute) character *entities*, and then parse the resulting text.
    >> So what gets passed from the XML parser to higher-level processors is
    >> e acute, not the five character sequence é. ...
    > I don't think you can be quite right, at least unless XML is quite
    > different from HTML here. For surely in both HTML and XML character
    > entities like &lt; can and should be used to replace the character "<"
    > when this is not to be interpreted as the start of a tag. This implies
    > that character entities are parsed not as the first stage of parsing,
    > but only after "<" is recognised as the start of a tag.

    I stand corrected. What I should have said is that an XML parser will
    first *replace* character entities, before passing the data to the
    consuming application. When that happens in relation to parsing (i.e.
    checking for well-formedness) is implementation-dependent, if I'm not
    mistaken. I find the XML spec a little fuzzy on that point (I can't
    wait for the English translation); it talks about at least &lt; and some
    other char entities being "escaped".


    This archive was generated by hypermail 2.1.5 : Wed Jul 13 2005 - 10:38:23 CDT