size of a byte and int?

In practice, though, such streams would almost always be files of 8-bit characters - even if the type happens to be wider. Of course, in theory, you're right, that could be a portability concern.

I was referring above to being unable to directly address 8 bit data (packed strings, or blocks read from such an 8 bit packed file), and requiring shift (or swap) and mask operations to do so.

Please desist from publishing these orphaned quoteless articles. There is no 'above' to which to refer. The google view of a newsgroup is sadly distorted from reality, as is their foul posting interface. See my sig below for a partial cure. Better yet, get a real newsreader. (Netscape, Thunderbird, Xnews all come to mind. Xnews is only for Windoze)

This is inadvertent - in addition to all their other sins, Google Groups is now incompatible with my habits. Thanks for the headsup, I'll try to remember to take the long route through their interface in future, which allows correct quoting. In this case the context of the posting was fairly obvious.

--Toby

There *was* at least one 1100-series C compiler. I never used it, but I know people who did. The compiler had 9-bit "char" (ASCII quaterword, as you would expect), 18-bit "short", and 36-bit "long". I think, but am not sure, that "int" was also 36 bits. All were of course ones' complement.

(This predated the 1989 C Standard, and I imagine various things the compiler did would not have been conforming. Note, however, that ones' complement is explicitly allowed.)