[Doc-SIG] Some random thoughts

[Laurence Tratt]

In message <LMBBIEIJKMPMLBONJMFCEEBECCAA.DavidA@ActiveState.com>
          "David Ascher" <DavidA@ActiveState.com> wrote:

> StructuredText has a couple of problems which we already discussed here.
> I agree with all of your comments about it, and I think there were a
> couple of other ones -- please check the archives for the details.

OK, I'm looking back in the archives (for various
lets-shut-down-the-mail-server-for-10-days-for-y2k, mailman set delivery off
for me on the doc-sig for about a month which meant I missed that)... Yeah,
OK. Although unless I've missed something there doesn't appear to be a
solution to the problem yet :(

> A few other points:
>
> 1) Does your version allow knowledge of the line numbers?  error-capturing
> is important, and StructuredText.py doesn't keep that.

Not yet. Currently there's no error handling (it's a bit too early days for
that).

Realistically, at the moment I reckon we need to sharpen up our ideas on
*what* we want, and then I'll recode from scratch based on that. At that
point I would probably put in full error handling and reporting. My current
implementation was less than a days work from start to finish, so I'm not
too emotionally attached to it <wink>.

> 2) The Zope folks will likely keep their own version as a separate fork
> until and unless your or anyone else's version does the same thing at
> least as fast -- they care about speed in Zope.

I doubt mine is faster, and I don't really care because it's not important
for my tasks (it's not API compatible with the current StructuredText
either); I can see that the current Jim module would probably be fairly fast
as it uses regex instead of re and some other little optimisation tricks. I
reckon it could be made even faster if they got rid of a lot of the small
function calls; maybe that's something on the cards. It'll reduce
readability even further though :)

> [I haven't had a chance to look at the code but will do ASAP]

Thanks, although bear in mind that it's not yet a masterpiece (in fact as
Peter Funk pointed out, in its undocumented state it's darn right confusing
at first). But it is generating some useful discussion which is good.


Laurie
-- 
http://eh.org/~laurie/comp/python/