Re: [tlug] wiki

["Stephen J. Turnbull" <stephen@example.com>]

bruno raoult writes:

 > Do everybody know SQL, and repair a database, knowing an index could be
 > wrong (after a crash for instance)?

Why does anyone need to know SQL?  All you need to know about SQL is
enough not to "emerge --unmerge PostgreSQL".  I agree that repairing a
database is a very specialized skill, but that kind of corruption is
very unlikely to happen (the ACID in databases is intended to prevent,
not cause, such corruption!  http://en.wikipedia.org/wiki/ACID)

What you're ignoring is the basic fact that every wiki depends on a
database system.  All wikis face race conditions because everything is
editable by anybody (in the traditional wiki configuration).  This
means that locking must be done.  A reasonable wiki will have
backlinks and things like that; those must be kept in separate indexes
(unless you're willing to let the wiki software edit the user's files
<shudder>).  Comments and stuff like that need to be associated with
appropriate pages.  I don't know what you call that; I call it a
database. :-)  As Walter Hansen points out, managing that in flat
files gets old fast.

Maybe the wiki's database is something other than an SQL database, but
it still needs to do "all that database stuff".  As long as that's
true, it makes a fair amount of sense to have the database stuff in a
database with a standard API.

Sure, I've read Jamie Zawinski's comments about "database worms" (in
http://www.jwz.org/doc/mailsum.html).  But in the context that he's
discussing (the single-task single-user application otherwise known as
a MUA), ACID is pretty easy to achieve.  (Note the last sentence of
his rant.)