Re: File position after fclose on a buffered input stream

[Neal H Walfield]

> > Nobody must make any such assumption and if there is such code it
> > has to be fixed.
> 
> I suggested that unbuffered input be used, would you agree that this
> is the best fix?

Upon further reflection, which code do you think is broken?  The
utility (in the aforementioned example, head) for "leaving" the file
position at then end of the buffered input (which according to SUSv3,
it is not required to worry about) or the shell code in the example?