Re: [Bug-gnupedia] Architecture Questions

[Mike Warren]

Rob Scott <address@hidden> writes:

> Well you're going to need a db anyway to manage things, and storing
> the entries in the db means that they are very easily searchable,
> the search is fast, and they are all in one place, as opposed to a
> large confusing directory structure.

For the classifiers, I can see this. For the backend, I can't. The
actual article store should be just that: a storage place for all the
articles.  There should be no searching, classification,
etcetera. These things should be separate and may indeed benefit from
a database implementation.

> If you're planning of just throwing loads of xml files into a
> directory, youre going to need a db to remember where everything is
> and what the titles are for searcing.

The ``titles'' are just the unique IDs, so no, the backend doesn't
have to search at all; the ``client'' (i.e. whatever is accessing the
database, which will almost certainly not be an actual user) just asks
for the article it wants by unique-ID in an easy manner:

http://www.gne.org/articles/unique-ID.xml

and the signatures, for the signature-happy folk ;)

http://www.gne.org/articles/unique-ID.xml.asc

> I dont know if any of what i just said makes sense, i just got up.
> But the main advantage is that this system in very clusterable,
> meaning that if we start to get more demand on the site, we just add
> more boxen.

Apache can load-balance, right?

-- 
address@hidden
<URL:http://www.mike-warren.com>
GPG: 0x579911BD :: 87F2 4D98 BDB0 0E90 EE2A  0CF9 1087 0884 5799 11BD