| Clark F Morris 2007-08-01, 6:55 pm |
| On Wed, 1 Aug 2007 18:10:53 +1200, "Pete Dashwood"
<dashwood@removethis.enternet.co.nz> wrote:
>
>
>"Richard" <riplin@Azonic.co.nz> wrote in message
>news:1185941885.429033.151800@e9g2000prf.googlegroups.com...
>
>I have no issue with any of the above. (Thanks for expressing it clearly
>and succinctly)
>
>However, it is predicated on certain assumptions (not unreasonable ones, but
>ones that most experienced programmers wouldn't do in code we wrote
>ourselves.)
>
> "If, for example, GO TOs are allowed from any point..." Would you write
>code like that? Do you know of any sites where standards permit code like
>that? So how likely are you to actually encounter code like that, in the
>real world? Yes, I know... you will encounter it occasionally, but it is far
>from the "norm".
>
>There has been a recurrent theme through this thread that just because the
>worst possible scenario COULD happen, it actually WILL. Well, maybe... but
>it certainly won't happen often enough to be a problem. When it does,
>there'll be an update to the site standards :-)
>
>Even when using ALTER, we didn't just allow uncontrolled open slather with
>GO TOs being tweaked all over the place. There has to be some discipline in
>place, otherwise chaos will rule whether you use GO TO, ALTER, SECTIONs or
>anything else...
You should have seen some of my payroll and marketing programs written
to fit in 16K to 24K partitions in the late 1960's and early 1970's.
PERFORM was simulated by using ALTER SWnnnn-FUNCTION TO
nnnn-FOLLOWING-PARAGRAPH. There were no PERFORMs. This was to save
memory when using the DOS360 compiler. Unfortunately some of the
marketing programs survived until at least the early 1990's. They
were efficient in terms of memory utilization but a horror from a
maintenance point of view. I did document all settings of each switch
which was identified with the letters SW but I hope that I would find
a better way of achieving the same goal.
>
>
>Pete.
|