Re: [avr-gcc-list] Standard IO - function fwrite

[Marjan Fojkar]

Joerg Wunsch wrote:
> (Please don't send HTML mails, it's cumbersome to handle.)
>
> Marjan Fojkar <address@hidden> wrote:
>
>   
> > Yes you are right. But if you use this function for binary data it
> > doesn't work.
> >     
>
> Why?
>
>   
> > Function fwrite enters into for loop and stays there
> > until putc returns zero or all data have been written.
> >     
>
> It's not putc() but rather the put() backend function you provided in
> fdevopen()/fdev_setup_stream(), and it stops when this put() function
> returns *non*-zero, as that indicates an error or EOF condition.
>
>   
I see now it is truly function put which represents pointer to my
backend function put and not putc. I'm gessing
it is time to buy glasses for myself :-)
And as you wrote below i have bug in my backend put function which 
behaves like putc.
Thanks for help!
> > If data contain zero values the fwrite will exit on first zero
> > value.
> >     
>
> I suspect you simply botched your put() backend function.  That
> function is supposed to return 0 on success, _FDEV_ERR (-1) on error,
> or _FDEV_EOF (-2) on EOF in case your driver/device can experience
> some kind of end-of-file situation that is distinguishable from a
> simple write error.
>   
> Btw., vfprintf() has a bug in that it ignores the error/EOF status
> when writing, and proceeds anyway.  Feel free to write a bug report
> for that.  Also, fputc() does not appropriately set the error and EOF
> flags, another bug.
>
>