Re: [Docutils-develop] CSV checkin

[David G. <go...@py...>]

David Priest wrote:
 > On Mon, 14 Jun 2004 20:37:50 -0700,
 > <doc...@li...> wrote:
 >> +``header`` : CSV data
...
 >
 > Commercial database output might very well use oddball CSV output
 > that is incompatible with ReST.  In particular, it may use line-ends
 > that aren't compatible with that used by the users' ReST,
 > *especially* if it's coming from a mainframe; or it may use field
 > separator or quote characters that are whitespace.

OK, then how about we settle on Excel format with
"skipinitialspace=True" for "header"?

Whitespace as field separators I can believe (TAB-separated data is
very common).  But whitespace as *quote* characters?  Do you have an
actual example of this?

 >> -``escape`` : [char | "tab" | "space" | "lf" | "ff" | "vt" | "hex ##"]
...
 >> +    .. We should be able to do without this one.
 >
 > Sweet Jesus, no.  Doubling-up CSV characters is apparently NOT an
 > uncommon  thing.

Re-read the checkin message.  Or better yet, do a CVS up (or check
<http://docutils.sf.net/sandbox/dpriest/csvtable/csvtable.txt> on the
web).  Diffs can be quite confusing; they aren't always minimal,
sensible, or obvious at first (or third) glance.

I did *not* delete the "escape" option, nor did I indicate that "We
should be able to do without" escape.  That comment refers to
"lineend", which it turns out we *can* do without (by reading in the
data and using .splitlines(); I tested all possible line endings to
prove to myself it wasn't necessary).

 > I also spotted something that might indicate that whitespace as
 > field delimiter has been removed.

Again, not the case.  I just added a couple of comments:

     .. Would whitespace other than tab or space ever be used?  I doubt
        it.

     .. Instead of "hex ##", use the same syntax as the "unicode"
        directive.

I don't think the complexity (implementation, documentation, and
conceptual) of "lf", "ff", etc. is worthwhile.  Rather than all that,
a simpler definition (char, character syntax as in the "unicode"
directive", plus "tab" & "space" just for delimiters) would be better.

-- 
David Goodger <http://python.net/~goodger>