Re: Closing Blueberry

[james anderson <james.anderson@setf.de>]

the "_really_ hard" reading creates two document classes, one of which
is judged non-well-formed because of the illegitimate presence of NEL's
within tags.

which also violates the the requirement, in that in one case character
data is returned, and in the other no data is returned.

which argues that the "_really_ hard" reading cannot be correct.

David Carlisle wrote:
> 
> > Read it more closely.  The *characters* (i.e. non-markup) returned
> > to the application are *as if* all CRs and CRLFs had been changed
> > to LFs before parsing.
> 
> I tried again, I still can't parse that paragraph the way you seem to
> have done.
> 
> It says it has to be as if the normalisation happened before parsing.
> I fail to see how anything that happens before parsing can distinguish
> between characters and markup. If the implementation did as suggested and
> normalised all line endings to #10 before parsing then any line end
> would be allowed (in the file) anywhere that #10 was allwed in the
> productions. I really can't see how this can be read any other way.
> 
> Actually if I try _really_ hard I can parse the paragraph as you intend,
> but it would never have occurred to me to do so, and I find it hard to
> believe that was the intention (but since there are people in this
> thread who were involved in producing that para, I leave it for others
> to judge what the original intention was)
> 
> David