Re: Types, byte codes etc.

On 5/7/07, Oren Ben-Kiki <[email protected]> wrote:
>
> On Mon, 2007-05-07 at 11:22 -0400, Brad Baxter wrote:
>
> > Small question: In this canonical format, how do you expect to treat
> > newlines that are in values, e.g.,
> >
> > [
> > "Romeo
> > Juliet"
> > ]
> >
> > or
> >
> > [
> > "Romeo\nJuliet"
> > ]
>
> The latter, for two reasons. First, I _think_ the 1st format is not
> valid in ECMA and JSON. Second, the second format does not raise sticky
> issues with regard to line break normalization.


Hi all,

I have another question, which may be better described as a point of
discussion.  I'm intrigued by this idea of the canonical format that
is essentially JSON with some YAML extras.  One aspect of
JSON (and btw of Perl's Data::Dumper format) is that a given object
may be represented on a single line, i.e., all newlines in the format
are represented by "\n".

So my question/point-of-discussion is: Could the newlines separating
tokens in the canonical structure be replaced with spaces and still
allow a YAML parser to correctly parse the structure?  Currently,
the answer is no for at least three reasons:

comments: # comment
directives: %YAML 1.2, %TAG !o! tag:ben-kiki.org,2000:
end of document markers: ...

There may be others, but these are the ones I found that seem to
rely on a newline to end them.

To answer yes to my question, I guess an extra form for
these could be introduced, for at least the canonical format.
I suppose the end of document marker--since it contains no
spaces--could be modified to recognize a following space as well
as a following newline.  The directives could perhaps accept a
quoted or parenthesized parameter list.  And perhaps a #' sort of
construct could say, "a comment follows that is delimited by a
single quote", e.g.

#' 'the ties that bind'

I'm not suggesting that these are the best options, just possible
examples.

One response is obviously, "Does anybody really need this?"

I don't know about "need", but I for one would not mind being able
to store serialized objects in a flat file, one object per line.  I can
do this with JSON (though they are simply scalars, sequences and
mappings--not true objects--and with Data::Dumper, including true
objects.   Of course, these do not have comments and directives
like YAML.

I guess the bottom line is simply this: Why not allow a YAML
stream, in its canonical form, to fit on a singe line? If it's not too
expensive--either technically or for readability--then one would at
least know that the option is there.

Again, I'm not expressing a need necessarily, but I have been
exploring flat file storage options. I think since a canonical format
is being formulated, this might be a good point to have this discussion.

Cheers,

-- Brad

-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/

_______________________________________________
Yaml-core mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/yaml-core