Xref: utzoo alt.folklore.computers:9138 comp.periphs:3404 Path: utzoo!utgpu!cs.utexas.edu!samsung!spool2.mu.edu!think.com!mintaka!spdcc!iecc!johnl From: johnl@iecc.cambridge.ma.us (John R. Levine) Newsgroups: alt.folklore.computers,comp.periphs Subject: Re: External Sorting Message-ID: <1991Jan30.015657.12840@iecc.cambridge.ma.us> Date: 30 Jan 91 01:56:57 GMT References: <1991Jan28.031017.19886@comp.vuw.ac.nz> <467@rufus.UUCP> Organization: I.E.C.C. Lines: 20 In article yossie@fnal.fnal.gov (Yossie Silverman) writes: >Does IBM still include a "READ BACKWARDS" instruction in their tape >drives? This instruction, I am told (and it seems logical) is primarily >for sort routines, where reading backwards on the tape could actually save >some time. Of course the bytes come in reversed, but that just needs some >extra code to deal with. I expect that start-stop drives still support read backwards. It's pretty hard to imagine that you can read streamers backwards, since they are used almost entirely for backup and logging anyway. By the way, the bytes from a read backwards do not come in reversed. You pass the channel the address of the end rather than the beginning of the buffer, and it knows to decrement rather than increment the buffer pointer during the progress of a read backwards. Gross, but quite effective. -- John R. Levine, IECC, POB 349, Cambridge MA 02238, +1 617 864 9650 johnl@iecc.cambridge.ma.us, {ima|spdcc|world}!iecc!johnl " #(ps,#(rs))' " - L. P. Deutsch and C. N. Mooers