Talk:Remote procedure call

From Wikipedia, the free encyclopedia

This article was originally based on material from the Free On-line Dictionary of Computing, which is licensed under the GFDL.

Contents

[edit] Questionable/Controversial RPC Definition

hi, I'm not sure whether I'd agree with the current definition that seems to restrict RPC to be only RPC if it's done remotely on a different *machine*, rather I think it generally considered accepted that remote procedure calls are also taking place if one process invokes calls/code in another process, regardless of whether that`s taking place on a different machine or not. I would have edited the corresponding paragraph, feel however that it would be appropriate to discuss my view before doing so.


You are quite right - I've updated the paragraph to show that rpc applies between two address spaces, which are commonly on two machines Peturbed 13:20, 2 January 2007 (UTC)

[edit] Windows - old flaws

Should have short mention of RPCSS attacks by W32.Gaobot and W32.RXBot http://www.symantec.com/avcenter/attack_sigs/s20386.html http://www.microsoft.com/technet/security/bulletin/ms04-012.mspx

Furthermore, I would like to know what, if any, common applications actually use RPC across a network.

[edit] include Java RMI? change definition?

shouldn't we include a reference to Java RMI? I think it belongs here as well.

Further, it says once we have object-oriented RPCs they are called remote method invocations. This implies that CORBA, which is object-oriented, should be called a rmi. Isn't that a bit confusing, as one usually uses the term for Java RMI only? I'd suggest we drop the object-orientation thing in the definition...

what do you think?

Seems that it's more about the implementors of Java RMI wanting to emphasize the object-ness of Java than any particular global naming strategy. I'd just add a note that RMI is a true RPC system, and perhaps something about the derivation of the RMI acronym.

[edit] No selection page

Why do we only put a small link to Rules of Professional Conduct and not an entry page like with most subjects where users have a choice?

Konerak 12:12, 28 December 2005 (UTC)

[edit] Web services

"Web services were the first real attempt to implement RPC between platforms."

  • WTF? How does DCE/CORBA not count as a "real" attempt? This doesn't seem like NPOV to me. MikeHearn 10:37, 25 January 2006 (UTC)
This section seems to be a mishmash of stuff, not all to do with web services, so I've tagged it for cleanup Keithdunwoody 01:37, 27 February 2006 (UTC)

[edit] RPC is synchronous

I think the article leaves alot for one's imagination, I read it without really understanding what it is! I got the following understanding of RPC from a book "Web Services - Concepts, Architectures and Applications" byt Gustavo Alonso and others.

[edit] History

In the evolution of information systems (going from a system all deployed in one mainframe, to systems deployed on several servers, see multi-tier architecture) RPC was the first and most simple way of calling procedures on other machines.

[edit] The RPC notion

To me the most important points when thinking of RPC is:

  • Calling a procedure on another machine (possibly even platform)
  • Doing it as if the procedure called was on the same system as the calling procedure
    • dont worry about network protocols when calling
    • don't worry about data formats (i.e. little vs big endian systems)
    • but you have to worry about them when setting up a system that enables RPC
  • RPC is a blocking (a more inprecise but also applicable term is "syncrounous") interaction. That is, the calling procedure will not continue before the called procedure returns.

--Velle 09:25, 3 May 2006 (UTC)

RPC is not necessarily synchronous[1]? Ewlyahoocom 10:52, 3 May 2006 (UTC)

[edit] Broken link

Following recent edits by user Mnot (talk ยท contribs), the Web service article no longer has a section named "Web services RPC", so the "Further information" link in this article is broken. I'll let one of the regular editors at that page fix this ;-). Cheers, CWC(talk) 17:54, 23 August 2006 (UTC)

[edit] Packet Format

Packet formats should be added

[edit] Updates

I've updated the page to remove some vandalism and clear up a few points. I moved the references to RPCGEN to a separate page as I thought this probably merited a page of its own.


The page is still slightly unclear/ doesnt read very well. ~Probably needs clearing up.

Should we put a reference to the 1984 Birrel and Nelson paper in the history section - it was an important work.

Peturbed 13:24, 2 January 2007 (UTC)


-This whole section could really use a rewrite, it's seriously underdeveloped. The Birrel paper would be a start, there's probably a few others we can add. Thomaslw 22:39, 21 February 2007 (UTC)

See our article on Bruce Jay Nelson for details of the Birrell and Nelson paper, including a link to a PDF. Cheers, CWC(talk) 05:04, 9 March 2007 (UTC)