Le mercredi 06 février 2008, Tom Lane a écrit :
> I'm really not seeing the case for user-level documentation of HOT,
> when for instance most of the planner's optimization behavior is not
> so documented.
In my POV, the case for HOT to appear in the documentation as been made by the
PR and Presskit. They both announce HOT to be a major performance addition in
8.3 that users will want to benefit from. Now we have to tell some more to
the users, I think we want to answer those two basic questions:
- How do I check that I'm using HOT? (you don't, transparent feature, blah)
- What's HOT is so good about compared to how PostgreSQL used to work?
My proposal was about answering this without requiring the user to be capable
of understanding internals and 'developer topics'.
If you want me to talk about another documentation missing entry, I'll be
happy to request a user targetted presentation of Executor Nodes, their input
and output, relative costs (simplified algorithms?) and the reason behind
their choice by the planner. I'm slowly becoming familiar with their names, a
little more each time I Explain [Analyze] a query, but not enough so to yet
be able to sketch an article about this...
I'd also like to add that people judging documentation effectiveness by its
length are certainly not real users of it. I've yet to meet a PostgreSQL user
who doesn't praise the product documentation, even if newcomers often need
some time to be able to understand where to find what. The Tsearch enabled
website has been a huge improvement here.
To ease PostgreSQL newbies manual grasping, I certainly don't think it will
need shortening the documentation.
Once more, all of this is only a enthusiastic user POV, the kind who think
he's helping when writing this mail ;)
--
dim