Code layout and debugging time

[Terry Lambert]

Jonathon McKitrick wrote:
> : So offseting discrete logic blocks that are intended to achive
> : specific goals makes it easier for the programmer to hold in
> : their head both the idea of what the code is intended to do, and
> : what their own logic dictates to them that the code actually
> : does.
> 
>  Do you feel your code does that, or are you one of the 'dense' style
>  programmers, subconsciously trying to avoid wasted space?

It depends on the problem being solved.  If it's not well
understood, then I will be very careful to make the code as
readable as possible.

If it's new code, I tend to document the crap out of it, on
general principles: I'd like it to be possible for someone
else to maintain, if necessary.  I also like my assumptions
going in and out to be known.  I tend to try for single-entry
and single-exit in most functions, which is about the only
way to deal with locking issues without introducing logic
bombs for the unwary who follow you.

For most things, I try to match the style of the existing
code in the area, so that there's not a radical change of
style in the middle of the code; nothing makes code more
unreadable than to change style in the middle of it.  It's
one of the reasons I disliked the ANSI C switch-over to
prototype-style declarations: you ended up with mixed code
styles in kernel sources, damaging readability.

I never put comments like "you are not expected to understand
this" in my code; if that were ever actually true, then I've
failed at my job (IMO).

-- Terry