Managing Google Groups headaches

Chris Angelico rosuav at gmail.com
Fri Dec 6 14:48:19 CET 2013


On Sat, Dec 7, 2013 at 12:32 AM, rusi <rustompmody at gmail.com> wrote:
> I guess we are using 'structured' in different ways.  All I am saying
> is that mediawiki which seems to present as html, actually stores its
> stuff as SQL -- nothing more or less structured than the schemas here:
> http://www.mediawiki.org/wiki/Manual:MediaWiki_architecture#Database_and_text_storage

Yeah, but the structure is all about the metadata. Ultimately, there's
one single text field containing the entire content as you would see
it in the page editor: wiki markup in straight text. MediaWiki uses an
SQL database to store that lump of text, but ultimately the
relationship is between wikitext and HTML, no SQL involvement.

Wiki markup is reasonable for text structuring. (Not for generic data
structuring, but it's decent for text.) Same with reStructuredText,
used for PEPs. An SQL database is a good way to store mappings of
"this key, this tuple of data" and retrieve them conveniently,
including (and this is the bit that's more complicated in a straight
Python dictionary) using any value out of the tuple as the key, and
(and this is where a dict *really* can't hack it) storing/retrieving
more data than fits in memory. The two are orthogonal. Your point is
better supported by wikitext than by SQL, here, except that there
aren't fifty other systems that parse and display wikitext. In fact,
what you're suggesting is a good argument for deprecating HTML email
in favour of RST email, and using docutils to render the result either
as HTML (for webmail users) or as some other format. And I wouldn't be
against that :) But good luck convincing the world that Microsoft
Outlook is doing the wrong thing.

ChrisA



More information about the Python-list mailing list