You don't know how lucky you are.

[achrist at easystreet.com]

Dennis Lee Bieber wrote:
>         I recall my college days (and no, I did have RPG available at my
> college -- BASIC, FORTRAN, COBOL, SNOBOL, APL were the HOLs) when
> someone taking the computer route at another college asked me for help
> on an assignment (we both were part-timers at Sears). HIS school used
> RPG.
> 
> ASSIGNMENT SUMMARY: You are given two sorted files. You are to produce
> one sorted file combining the two.
> 
>         His planned solution... Copy F1 to Fout, append F2 to Fout, invoke
> system sort on Fout.
> 
>         Apparently his school failed to cover any moderate algorithm design...
> Like the common sort/merge algorithm (and since he already had sorted
> files that problem just reduced to just the merge part).
> 
>         Footnote: His school also managed to place him in a job after
> graduation -- I have no further news of his eventual end...

Not always wrong.  I did some timings on big IBM machines, and the
sort programs on those things are very highly optimized.  I found
that the sort program (syncsort, IIRC) could read + sort + write a
file faster than I could simply copy it with PL1 code (optimizer
on, record IO, properly blocked).  Of course, a decent sort program 
could be set up to do a merge, so you could save some of the copying,
but I believe that it would also detect when the input consisted of
just 2 runs (or maybe 1) and just do the merge.  And by requesting the 
redundant sorting, you would be completely protected from any need
to worry about any crash or hang-up or bad output if someone
screwed up and gave you a file that was unsorted, or sorted on
the wrong fields, or sorted using a different collating sequence.
WTH wants to program all that error-handling?  

The XP guys always ask, "What's the simplest thing that could 
possibly work?"  Sounds like he found it.  RPG guys were kind of
like that.  I had one working for me for a while, and I couldn't
keep him out of my office.  Everything I gave him to do, he'd
be back in 15 minutes with it finished.  I don't think he knew
much about alogorithm design,  and I didn't want to know what
was in his programs any more than I wanted to know what was in
my lunch at the company cafeteria,  but whatever it was, no matter
how ugly, he was good at it.


Al