Very Computer

In a program I just finished I am using strtok to extract TCP messages
one at a time ( with a DELimeter character set to ASCII 10 line feed )..
 If you call strtok the first time with a NULL argument
i.e like this strtok ( NULL, (char *) &DEL );
then it returns NULL which suits me fine.... but I'm just curious if
this is the defined behaviour....d

 More importantly, in a program that does read()s ( over a TCP
connection ) and which also uses signals I am using code which looks
like
 while ( read(...) == -1 )
 { if ( errno != EINTR ) { /* unrecovable error */ }
    /* just a signal... try again */
 }

 ( the read call is blocking )
 This doesn't work if read() is interrupted while reading .. it will
simply return with the number of bytes it managed to read... What are
the solutions to this rather fundemental problem?
/alex

p.s What is the cause of an IOT / Abort signal?

-Rob Ryan

Quote:>  More importantly, in a program that does read()s ( over a TCP
> connection ) and which also uses signals I am using code which looks
> like
>  while ( read(...) == -1 )
>  { if ( errno != EINTR ) { /* unrecovable error */ }
>     /* just a signal... try again */
>  }

>  ( the read call is blocking )
>  This doesn't work if read() is interrupted while reading .. it will
> simply return with the number of bytes it managed to read... What are
> the solutions to this rather fundemental problem?

ron

--

 \ WWW: http://www.tu-chemnitz.de/~row/            \

   \ PGP key available by finger to my email address \

Quote:> In a program I just finished I am using strtok to extract TCP messages
> one at a time ( with a DELimeter character set to ASCII 10 line feed )..
>  If you call strtok the first time with a NULL argument
> i.e like this strtok ( NULL, (char *) &DEL );
> then it returns NULL which suits me fine.... but I'm just curious if
> this is the defined behaviour....d

BTW, if you really expect strtok to work you need to pass a _string_
as the second argument, not a pointer to a single char (as seems to be
the case).  As written, that will work only due to luck, or perhaps to
a compiler that zero-fills padding inserted between variables for
alignment (whcih means it could break if you change the order of
variable declararations, or compile with different options, or any
damn thing).