Where should the type information be: in tags and descriptors 434
Still fighting the DECsystem 10-20 versus VAX battles?
RMS - the Record Debt Management System - doesn't particularly care what is in the file, unless it is told to impose some particular structure on it. If you want to blame something, blame the tools that do the transfer for poor buttumptions (and I'm guessing that the tools you were using were 10-20 years ago written for VMS to VMS transfers). Frankly, RMS isn't going to change the bits in the file on a whim - it will likely set the record or file attributes incorrectly, and then when asked to pbutt it along to the next non-VMS node, the transfer utility may end up sending something unexpected by using the wrong file semantics.
If I transfer a RMS file from a VMS system through a Windows or UNIX system using, say, FTP - the same things happens. First I need to tell it if it is an ASCII or BINARY file, and then on the other end (buttuming I again transfer it back to a VMS system) - I probably will have to correct the file attributes because of the fact there is no way to describe to the windows or UNIX system attributes that it doesn't understand. And if I don't tell FTP that it is a BINARY file, it *will* screw with the bits.
FWIW - RMS isn't a file system. Its a library interface designed to allow the imposition of structure on a file. The file system for VMS is ODS-2-5 - which just knows about blocks or clusters of blocks. Most utilities and languages on VMS access the file system through RMS services - including the various utilities to transfer data. Some are better than others for doing non VMS-to-VMS transfers, but especially 25 years ago, the
Where should the type information be: in tags and descriptors 436
B ? Why B instead of U ? +--------------- I forgot. (It *was* back in 1965, after all.) I *think* it had something to do with the kind...
Your argument on VMS not being a "timesharing" system and being "task oriented" is also -- well a bit misleading. VMS is used widely as a timesharing system. You confuse underpowered VAX 780's versus a then contemporary Tops10-20 system as a system misdesign. VMS doesn't marshall all the resources needed for a "task" before starting it. It does provide controls for how resources are allocated, and it can be tuned to be realtime-like versus a more clbuttic timesharing system. I still work on a cluster of VMS systems that can at any time have hundreds or thousands of logged on users. Remember that VMS could be used for real time use, or for effectively single-user technical work, as a batch oriented system, or as a general purpose timesharing system - depending on how you want to tune it, and what type of policies you want to set.
Perhaps that flexibility means that it isn't the "best" at any one of these things (it can be "near" real time for example) - but it meant that you could use it for many purposes.
FWIW - I started at DEC about 26 years ago from the other end - working on PDP-11's, RSTS, and RSX11M in the field. I don't know if killing the 10-20's was a good idea or not, but that doesn't mean that the VAX or VMS wasn't a good idea.
Where should the type information be: in tags and descriptors 439
Want the long list or the short one? Here are some that I know about for certain, various time frames from 1960 to 1980. DOS, DOS, CTOS, NOS, HDOS ... and at least a dozen...
Where should the type information be: in tags and descriptors 435
Sigh! I'm not talking about a file on the system. I'm talking about a file system that is...
Alt Folklore Computers Newsgroups