ddb@mrvax.DEC (DAVID DYER-BENNET MRO1-2/L14 DTN 231-4076) (02/18/85)
>I don't quite understand (nor do I know what this discussion is doing in >net.arch). Why should every program need to understand indexed files? >All you need to do is pipe the output of the indexed-file reader into the >other program. When I open a file for sequential access in a program, I don't want to have to think about whether the program could also be accessed some other way. With RMS, I don't need to. With less-integrated systems, I do. When I TYPE a file, I don't want to have to think about what other modes of access might or might not work for that file; I shouldn't have to remember that the file is indexed, remember the name of the un-indexer, and construct a command line to pipe the file through the un-indexer. Now, I STILL don't like RMS; but a standardized file-system interface is very convenient, it's just some of the particular implementation choices that I object to. >What, you say you don't have pipes in VMS? Maybe you'd better put them >into RMS. :-) >-- >In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 4251) Actually, pipes should probably be done as a device driver for a virtual device. I imagine it's been done many times. -- David Dyer-Bennet -- ...decwrl!dec-rhea!dec-mrvax!ddb
guy@rlgvax.UUCP (Guy Harris) (02/20/85)
> When I open a file for sequential access in a program, I don't want to have > to think about whether the program could also be accessed some other way. > With RMS, I don't need to. With less-integrated systems, I do. > When I TYPE a file, I don't want to have to think about what other modes > of access might or might not work for that file; I shouldn't have to > remember that the file is indexed, remember the name of the un-indexer, and > construct a command line to pipe the file through the un-indexer. But how many indexed files are TYPEable (or "cat"able, or whateverable)? Most of them probably contain lots of binary data in their records, which means you'll get gibberish if you type them - if you're lucky. > >What, you say you don't have pipes in VMS? Maybe you'd better put them > >into RMS. :-) > Actually, pipes should probably be done as a device driver for a > virtual device. I imagine it's been done many times. What's wrong with just using mailboxes? A mailbox has somebody who stuffs records into them and somebody who pulls records out of them; I believe you can even open one as an RMS file and pump text (or any other sequential record format) into one or pull it out of one. That would be sufficient to qualify them as pipes in my mind. I don't think DCL has any syntax to set up pipelines, however. The VMS/VNX/whatever Bourne shell might. Guy Harris {seismo,ihnp4,allegra}!rlgvax!guy
jlg@lanl.ARPA (02/22/85)
>(nor do I know what this discussion is doing in >net.arch). This discussion is on net.arch because it is about the archetecture of operating systems. The assumption that 'archetecture' refers only to the silicon is overly restrictive. If there were a net.os.theory, that's probably where this discussion belongs. J. Giles