Where should the type information be 165

snip

Yes, VSE supported 3370. I don't know about 3310.

Yes, I know. The comment above was, as stated, about how non-IBM compatible mainframes, with fixed length disk blocks handled record oriented I-O where the blocks were read directly into a user buffer instead of a Unix style centralized cache.

Thanks for the clarification. We may be talking at cross purpose here. I am saying that the mechansim I described was very common on mainframe systems. I take your comment to say that minis of the day didn't have the software, or perhaps the memory to do that. I can certainly accept that, but it doesn't affect my point that such mechanisms were in common use prior to the Unix mindset taking hold.

Yes, as far as the term is concerned (except that it wasn't an HLL term, as it was a parameter to the DCB macro), but that is why I said "functionality equivalent to". As to the rest of your paragraph, I agree wholeheartedly. I was arguing against the position that no one used such mechanisms. As you point out, it was *the* mechanism used by most, if not all, mainframe systems. I think we are "in violent agreement" here! :-)

-- - Stephen Fuld e-mail address disguised to prevent spam