Re: [xml-dev] Quiz: is this XML well-formed?

[Marcus Reichardt <u123724@gmail.com>]

Thanks Michael Sperberg-McQueen for sharing that bit of XML/ERB history.

Digging a bit deeper into old posts to this very mailing list, I found
that various whitespace issues were discussed at great length in
http://lists.xml.org/archives/xml-dev/199708/threads.html .
However, these seem to be about whitespace in content rather than
attribute specifications, and in particular about a replacement for
SGML's traditional newline (RE/RS) handling rules consistent with
XML's DTD-lessness/-neutrality (remember traditional SGML considers
whitespace around start-/end-tag elements as non-significant "layout"
if appearing all on a line of it's own, for declared elements ie not
those included via content exceptions). And also, in fact convenience
for line-oriented Unix tools and Perl was hotly debated.

Other than that, the earliest mention of the issue on this mailing
list appears to be in
http://lists.xml.org/archives/xml-dev/199703/msg00083.html where the
relevant BNF grammar rule reads

    Markup ::= '<' Name (S Name Eq QuotedCData)* S? '>'   /* start-tags  */
                 ...

So my guess would be that the requirement for space before an
attribute name simply slipped into a first draft for XML because it
was more compact and unambiguous than, say

    Markup ::= '<' Name (S (Name Eq QuotedCData S?))* S? '>'

and was never much of a topic for discussion thereafter. Notably, it
isn't mentioned in James Clark's "Comparison of SGML and XML"
(http://www.jclark.com/xml/NOTE-sgml-xml.html).

Note that SGML also allows

   <test att =  "x" otheratt = "y">

(ie with spaces before/after "=") and at least libxml2 (xmllint)
happily accepts that syntax without complaining.

From the same era, I'd also like to share Paul Prescott's answer to
"Must XML be SGML-compatible?", as timely as ever:
http://lists.xml.org/archives/xml-dev/199710/msg00189.html

Regards,
M. Reichardt
sgmljs.net