Actually, that's not obvious at all :-) My prototype code for MaVerick's dynamic files (and no, that code is not part of Rob's MaVerick code) would never have run out on a delete.
Let's say I was merging two buckets - 5 and 13 for the sake of argument. If 5 was the fullest I just reset the pointers so 13 was now an overflow bucket for 5. If 13 was the fullest, that got reset to be the new 5, with the old 5 as overflow. Then further code packed the group, moving records from overflow buckets into the primary bucket as appropriate. So no - you could never get an overflow on delete there. Obviously, you know much more about the internals of UV than I do, but you're presuming that other peoples knowledge of the internals is as good as yours. Not a sensible presumption ... :-) Cheers, Wol -----Original Message----- From: Martin Phillips [mailto:[EMAIL PROTECTED] Sent: 29 May 2007 10:09 To: [email protected] Subject: Re: [U2] [BB] U2 Enhancement Request - WRITE What is less obvious to most developers is that ON ERROR might just be useful on a DELETE too as it is technically possible to run out of disk space while deleting a record. (How? Think about a dynamic file where the delete causes a merge that results in overflow). Martin Phillips Ladybridge Systems Ltd 17b Coldstream Lane, Hardingstone, Northampton, NN4 6DB +44-(0)1604-709200 ------- u2-users mailing list [email protected] To unsubscribe please visit http://listserver.u2ug.org/ ------- u2-users mailing list [email protected] To unsubscribe please visit http://listserver.u2ug.org/
