home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!wupost!cs.utexas.edu!swrinde!sdd.hp.com!network.ucsd.edu!mvb.saic.com!macro32
- From: raxco!galaxy.dnet!gleeve@uunet.UU.NET
- Newsgroups: vmsnet.internals
- Subject: Moving DDTs etc.
- Message-ID: <9209091400.AA19886@relay1.UU.NET>
- Date: 9 Sep 92 13:22:19 GMT
- Organization: Macro32<==>Vmsnet.Internals Gateway
- Lines: 32
- X-Gateway-Source-Info: Mailing List
-
- I'm inclined to agree with Jon Pinkley's extensions to my suggestions
- about moving DDTs into intercept UCBs. (And yes, I meant DDTs, not
- DDBs...too many of these darn three letter acronyms occasionally
- fry my brain when posting...)
- While such suggestions cannot be enforced, having a number of folks
- publically promise to follow them (and I will in some code I'm developing
- for starters) will help to create support for the standard. In the
- case where nothing else is using the method, no problems will ensue.
- If we have a case where two programs try to steal the same resources,
- we have currently no way to interoperate and have only the options of
- refusing to work at all, or of crashing a system. Proposing and living
- with a standard gives all developers a third and better alternative.
- The fact that some will not get the word, ever or at least not for a
- while, is not a reason not to begin. I will store the messages for
- the sig tapes and hope some of us can get some intercept code out
- there also.
- As for Alpha drivers, I have too little information, but suspect
- something like this is desirable. In a new design I'd ask for some
- customer queue headers in UCBs, in the exec, and in IRPs that would
- be usable for add-ons and which would have some sort of identifier
- scheme written into a standard so they could be used by multiple,
- relatively uncoordinated designers. (By "relatively uncoordinated"
- I mean that identifiers might collide but the probability of this is
- low.) The user SYSGEN parameters in the current VMS are an interesting
- start but are too risky to use for purposes other than for one
- specific site...too likely they'll be clobbered by someone else's
- code/data. Someplace to hold kernel tables that's faster to find than
- a lock value block or logical would however be a win.
- For altering driver flow, I'd say that the present suggestions
- look very good and we should spread them around.
- Glenn
- Everhart@Raxco.com
-