home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.client-server:176 comp.unix.osf.misc:22 comp.os.ms-windows.programmer.win32:2044
- Path: sparky!uunet!ogicse!das-news.harvard.edu!husc-news.harvard.edu!husc-news!jbaker
- From: jbaker@husc8.harvard.edu (James Baker)
- Newsgroups: comp.client-server,comp.unix.osf.misc,comp.os.ms-windows.programmer.win32
- Subject: dce rpc -- who supports it? and other thougthts
- Keywords: NT RPC DCE OSF/1 calling protocol
- Message-ID: <JBAKER.92Nov18215657@husc8.harvard.edu>
- Date: 19 Nov 92 05:56:57 GMT
- Article-I.D.: husc8.JBAKER.92Nov18215657
- Distribution: comp.client-server
- Organization: Harvard Arts and Sciences Computer Services, Cambridge, MA
- Lines: 41
- Nntp-Posting-Host: husc8.harvard.edu
-
-
- My main question is: who supports DCE RPC now? And for the other
- players, when? So I know Window NT does -- I have played with it --
- and I guess OSF/1 which DEC supports should. But will IBM support it
- for its mainframes and on what: VM/GCS? Any wild ideas are appreciazted.
-
- As for my experience with DCE RPC on Windows NT, it seems to be a
- reasonable implementation of the RPC paradigm, as I understand it from
- reading Power Programming and RPC and some other less useful books.
- Not that I don't have gripes. I would like to be able to control the
- various queues with my own service discipline, or at least not without
- altering the stub code. An example would be to allow asynchronous
- calls and other less generic calling protocols. Of course, ONC RPC
- has a rather simple way of supporting an asynchronous protocol --
- zero timeouts -- but could those fail on a machine with both client
- and server on the same machine? DCE RPC, on the other hand disallows
- this trickery by making a zero timeout something else: a qualitative
- minimum time that a response should take. The attitude that DCE RPC
- takes is either you are going to spin a thread off to call a long
- running RPC routine or you are going to break it up into two and spin
- to find when the results are ready. I am sure it is indeed for the
- best ;-).
-
- In the article in Microsoft Systems Journal about the NT architecture,
- Helen Custer (?) mentioned that the client-server support for local
- (same machine, different address space) RPC is optimized by thread
- pairing between the client/server. Does MIDL/ACF support this with
- the localrpc protocol, currently so well-documented ;-)?
-
- BTW, am I right in thinking that OSF is some secret society? I bike
- by their place all the time in Cambridge -- and strangely I never hear
- of them doing anything except in the pages of ComputerWorld and PC
- Week and the Burton Group. Nor is there a great abundance of postings
- in comp.unix.osf. Do I have to be on some mailing list?
-
- Thanks.
-
- -- Jim
-
- jbaker@husc.harvard.edu (forwarded to ...)
- jimb@kenan.com
-