Re: Benefit of not defining the order of execution

Kaz Kylheku <kkylheku@xxxxxxxxx> wrote:

On 2009-02-13, Stephen Sprunk <stephen@xxxxxxxxxx> wrote:
If the order actually matters, you can
modify the code slightly

Here is 250,000 lines of code. Go find places where the order matters and fix
the code ``slightly'' so that it actualy expresses what you think its
author wanted it to do!

Look here, boy, if you wrote 250,000 lines of code without a fucking
clue about the language you wrote them in, that's _your_ problem,
Buster, not mine.

All done? Great, let's recompile and ship the new version to the customers!

_Your_ customers. Mine are used to getting more solidly written code
than that. Yours, apparently, won't give a ***.

Here is a quote:

Having the order of evaluation undefined is claimed to yield better
performing code. Compilers could warn about such examples, which are
typically subtle bugs (or potential subtle bugs). I'm disappointed that after
decades, most compilers still don't warn, leaving that job to specialized,
separate, and underused tools.

Who wrote this?

It doesn't matter, because its meaning is muddled and its concept is
wrong from the start. "Such examples"? _Which_ examples, exactly, ones
which depend on the order of evaluation or ones which (correctly!) do
not?

Richard
.