Path: csiph.com!fu-berlin.de!uni-berlin.de!individual.net!not-for-mail From: rbowman Newsgroups: comp.os.linux.misc Subject: Re: VMS Date: 1 Jul 2025 04:02:55 GMT Lines: 28 Message-ID: References: <4_GdncCsf-Nqe8n1nZ2dnZfqnPSdnZ2d@giganews.com> <103392c$lpbg$5@dont-email.me> <1033o4a$1qj6$3@dont-email.me> <1033tv1$3aqu$3@dont-email.me> <1034pj8$a74s$1@dont-email.me> <103teq1$220b9$1@dont-email.me> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit X-Trace: individual.net Lc4n0g6X3iRWkgmDh1PRsgvElFJpYkvsaBj7OwwNnGdwXnn6+u Cancel-Lock: sha1:sNF9iqD1eK0MdRhRdSo28/ZF0Lo= sha256:J43OA/DiAPymixgZUjpaYjfpoKc8gac1G0CZxivKagg= User-Agent: Pan/0.160 (Toresk; ) Xref: csiph.com comp.os.linux.misc:69239 On Mon, 30 Jun 2025 23:12:21 -0400, c186282 wrote: > MOST compilers, I think ints are still almost always 16-bits as a > holdover from the good old days. You can declare long and long long > ints of course, but int alone, expect it to be 16-bit. Not any compiler I've worked with in the last few decades. That sort of went out with CP/M. Way back when bytes were worth their weight in gold someone declared 'short ObjNum;' That's rather important since that is the number of objects that can be handled by the system including incidents, comments, persons, vehicles, alerts and so forth. Being signed and 16 bits the maximum value is 32767 It got by for an amazingly long time but as larger, busier sites came along the system ran out of object number. It wasn't pretty. Edit a number of files to make it an unsigned short and you get 65535. It was close a couple of times but with some sophisticated reuse strategies there never was a disaster. Why not make it an int? Even with a signed int 2147483647 wouldn't be a problem. Because an int is 32 bits. Every struct, every XDR encoding, database, and so forth would have to be modified so we crossed our fingers. In DB2 SMALLINT is 16 bits, INT is 32 bits. SQL Server is the same. Both of them use BIGINT for 64 bits.