Is there anything in c standard about the operation of file streams etc. in
reguards to how they are supposed to act if an over flow occurs. I am trying
to work out the correct way to handle files that are greater than 2^32
bytes. Different compilers/operating systems etc. all behave different for
ftell. I guess the answer is its undefined...

"MisterE" <MisterE@nimga.com> wrote:

> Is there anything in c standard about the operation of file streams etc. in
> reguards to how they are supposed to act if an over flow occurs.

No. Except, of course, that merely writing to a file which is about to
"overflow" (quotes because it's not the same thing as a real numeric
overflow) may continue to write properly; may write garbage (since no
conforming program can detect the difference anyway); may return with an
error status as it would for, say, a file without write access; but may
_not_ crash the program, return nonsense, or exhibit other truly
undefined behaviour.

Richard

MisterE wrote:

> Is there anything in c standard about the operation of file streams
> etc. in reguards to how they are supposed to act if an over flow
> occurs.

Overflow for a file?

All that Standard C guarantees is that the various I/O functions of the
Standard library, (like putc, printf etc.), will return a status value
indicating success or failure. The variable `errno` may or may not be
set to any meaningful value.

Thus though you can detect a failed operation, it's generally difficult
or not possible with Standard C to find out _why_ the operation has
failed. You might have to depend upon implementation and system
specific methods.

> I am trying to work out the correct way to handle files that
> are greater than 2^32 bytes.

On 32 bit systems, you very probably have to use non-Standard
alternatives like ftello, ftello64 etc.

You shouldn't have a problem on 64 bit systems.

> Different compilers/operating systems
> etc. all behave different for ftell.

No. ftell behaves the way the Standard defines it. However it may or may
not be sufficient under certain conditions.

> I guess the answer is its
> undefined...

It _is_ defined. You're going beyond it's specification however, so what
do you expect?

On Oct 22, 7:37 am, santosh <santosh....@gmail.com> wrote:
> MisterE wrote:
> > I am trying to work out the correct way to handle files that
> > are greater than 2^32 bytes.
>
> On 32 bit systems, you very probably have to use non-Standard
> alternatives like ftello, ftello64 etc.
>
> You shouldn't have a problem on 64 bit systems.

Isn't this at least part of the reason fgetpos() and fsetpos() exist?
fpos_t is allowed to be larger than a long, and even a non-integral
type, to handle possibly huge files.

Justin Spahr-Summers wrote:

> On Oct 22, 7:37 am, santosh <santosh....@gmail.com> wrote:
>> MisterE wrote:
>> > I am trying to work out the correct way to handle files that
>> > are greater than 2^32 bytes.
>>
>> On 32 bit systems, you very probably have to use non-Standard
>> alternatives like ftello, ftello64 etc.
>>
>> You shouldn't have a problem on 64 bit systems.
>
> Isn't this at least part of the reason fgetpos() and fsetpos() exist?
> fpos_t is allowed to be larger than a long, and even a non-integral
> type, to handle possibly huge files.

Yes. I should have mentioned that. When the OP said he was using ftell,
(and encountering problems), I automatically thought of ftell-like
alternatives.

As you said, fgetpos and fsetpos are Standardised and better.