Re:Subtleties of the ANSI/ISO C standard [pdf]
by wuch on Wednesday December 23, 2015 @03:12AM
(#10782435)
What authors point out is not trivial observation that programs are running on
finite-memory machines and thus cannot be Turing complete. What they do point
out is that even if C programs have been running on infinite-memory machines
the memory accessible to C programs is still bounded.
In general I would agree with authors on this point as long as we exclude
external storage. They also hint that, while it would be trivial to provide
interface to infinite memory tape, the standardized part of file I/O does not
seem to do that. They claim that it is prohibited by bounded return value of
ftell. I would disagree, because ftell may return an error value in this
scenario as well. Moreover to access infinite tape you don't even need ftell,
relative movement provided fseek suffices.
In general I would agree with authors on this point as long as we exclude external storage. They also hint that, while it would be trivial to provide interface to infinite memory tape, the standardized part of file I/O does not seem to do that. They claim that it is prohibited by bounded return value of ftell. I would disagree, because ftell may return an error value in this scenario as well. Moreover to access infinite tape you don't even need ftell, relative movement provided fseek suffices.