🚀 go-pugleaf

Ian Cargill (ian@soliton.demon.co.uk) wrote:
: It has been asserted elsewhere by someone that:

: > "The only guarantee that C gives is that the following data 
: >  structure can be written one way and read another."  
: >
: >   union {
: >     unsigned char a[4];
: >     int b;
: >     }

: I replied with my opinion that this is actually implementation defined 
: behavior, and that the only guarantee the standard gives is regarding 
: access to common initial sequences of structs in a union.

: Since there is still some dispute, I would welcome comment from the 
: assembled multitude, particularly those who know what they are talking 
: about.  A majority verdict will do. :-)


I would not like to say it even gives that guarantee. I think you can
say you can write .b and read .a (though what is read is useless). I am 
not sure that you can write .a and read .b as I am not sure if it is
guaranteed that all combinations of the bits of the machine representation
are legitimate integer values or if some can lead to undefined behaviour 
(eg a parity check failure). (It is the case that not all the bits have to 
be used as part of the integer value.) However I do recall some argument that 
all bit combinations must give as defined behaviour (though implementation 
defined) as a _valid_ integer. In this case you can read and write either
way (though neither is useful).

If you changed a to be a[sizeof(int)] then you could (I think) guarantee that
copying to another variable of the union type then you would have copied the
value of b too.

--
Stephen Baynes                              baynes@mulsoc2.serigate.philips.nl
Philips Semiconductors Ltd
Southampton                                 My views are my own.
United Kingdom