RE: Safe to maintain pointer to errno() ?

Thanks, Scott. And Christen and Jon and Barbara!

In fact, I have getErrno and setErrno in a standard service program. (I'll
review that errno-trap thingy from Christen - the name is interesting!)

My real reason for asking was that I was considering changing it to only get
the pointer one time.

W.r.t the "why fight it" question, I'm inclined to agree. But since some of
our jobs process a few million rows at a time, and for each row there can be
significant errno testing, the insignificance of the pointer retrieval might
add up to measurable values.

But I'll pay my toll, as I would without question in c. It was a point of
curiosity and a potential opportunity for minor gain.



Dennis Lovelady
http://www.linkedin.com/in/dennislovelady
--
"Advertising may be described as the science of arresting the human
intelligence long enough to get money from it."
-- Stephen Leacock

My specific question is whether it is safe to obtain a pointer to
errno only one time, and then reference the based value from then on.

When I asked IBM this question, I was told that the pointer probably
won't change, it'll probably always point to the same area in memory.
but they won't guarantee that this will always be the case, so they
strongly recommend that you retrieve the pointer every time. (As ILE C
programs do.)

Since the overhead of retrieving the pointer is insignificant, I just
meekly complied.

It makes sense to me that errno's value would be at a static location
and won't float around in memory, but - again - I haven't seen any

guarantee

of that.

In all of my tests and research, it is indeed a static location that
never changes. The only problem is it's not guaranteed to _always_ be
that way, going forward. So you risk compatibility breaking at some
random time.

In C they do this:

extern int *__errno(void);
#define errno (*__errno())

So every time you reference errno in C, it does call the __errno()
function, and the OS could potentially change the pointer. If it does,
your C code will keep working...

So IMHO, RPG code should work the same way. It's easy enough to write
subprocedures named geterrno() and seterrno() that retrieve the pointer,
and then return/set errno appropriately.

So why fight it?

--
This is the RPG programming on the IBM i / System i (RPG400-L) mailing
list
To post a message email: RPG400-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/rpg400-l
or email: RPG400-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at http://archive.midrange.com/rpg400-l.