Re: Avoiding disk arms with CLOB

On Thu, Sep 19, 2013 at 2:52 PM, Rory Hewitt <rory.hewitt@xxxxxxxxx> wrote:

This doesn't mean that you shouldn't *think* about performance up-front.

It's not so much that you're not supposed to think about performance;
rather, you're supposed to focus on (1) correctness and (2) clarity.
If you've got reasonable programming skills, keeping your focus on
these things will typically already result in reasonably performing
code, and high maintainability.

It's where chasing performance conflicts with "the obvious way to do
it" that you have to beware, and really check to see that you've got a
performance problem at all, and if you do, then where.

For instance, in RPG programs, [the bottleneck is] *
probably* going to be I/O-related or due to repeated memory
(re)allocations or 'communications' (especially if web- or
socket-related) and not in the code itself, unless you're
doing crazy bad loops...

There's almost nothing *to* code in RPG except for I/O, memory
(re)allocation, or communications. What usually *defines* a "crazy
bad loop" in RPG is doing unnecessary repetitions of I/O, memory
(re)allocation, or communication.

Too often, Knuth et al. are invoked and result in bad code being written,
on the theory of "Let's get it done and then worry about making it fast".
And then there's no time/money to fix any performance-related issues...

I understand and appreciate what you're saying, but honestly, I think
most bad code results just from not thinking enough, period. (As
opposed to not thinking enough about performance.) When I see bad
code, it certainly doesn't look like the person who wrote it was
thinking "I'm not sure how fast this code will run, but Knuth says I
shouldn't worry about it, so I'll leave it the way it is".

John