home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!darwin.sura.net!Sirius.dfn.de!zam103!djukfa11!hac041
- From: HAC041@DJUKFA11.BITNET (Konrad Hinsen)
- Newsgroups: comp.sys.atari.st.tech
- Subject: Re: XAcc protocol definition
- Message-ID: <92357.133157HAC041@DJUKFA11.BITNET>
- Date: 22 Dec 92 13:31:57 GMT
- References: <92356.104533HAC041@DJUKFA11.BITNET>
- <1992Dec21.205804.19515@netcom.com>
- Organization: Forschungszentrum Juelich
- Lines: 48
-
- In article <1992Dec21.205804.19515@netcom.com>, ersmith@netcom.com (Eric R.
- Smith) says:
- >
- >The biggest problem that I see with the XACC protocol is the requirement
- >that data be passed in pointers; that's not a nice thing under MultiTOS,
-
- I understand your dislike for pointers, but there are good
- reasons for them:
-
- - it is not possible to send data objects of arbitrary size using
- AES messages
- - it is more efficient to pass only pointers
-
- Besides pointers are also used in other places (XBRA), so MultiTOS
- will have to provide a means for pointers across processes anyway.
-
- >since it means that memory protection must be bypassed (with some care
- >one can restrict the memory protection bypassing to a specific memory
- >block, but I'm afraid that many programmers won't take this care). It
-
- I have a bit more faith in application programmers, but you might be
- right. Perhaps this should be pointed out more explicitly in the
- documentation. At least I can assure you that all programs currently
- using this protocol allocate a buffer for the data to be passed
- by Mxalloc() without allowing access to other data.
-
- >also makes any kind of networking difficult, although admittedly that's
- >a less serious concern at the moment since the AES doesn't support
- >networking.
-
- Networking would require a complete new set of AES functions, so
- obviously there is no way to define data interchange protocols
- now in such a way that they would work unchanges with future
- networking versions.
- >
- >In MultiTOS there will be a drag and drop protocol specified, using
- >AES messages and MiNT pipes for communication.
- >
- Really? It has been announced for quite some time, but so far I have
- not seen anything more specific.
-
- Anyway, this is a somewhat different problem. While drag&drop could
- be implemented using XAcc, the reverse is obviously not true.
- Besides XAcc was designed for standard TOS and can still be used there,
- which explains why it can't use pipes and other MiNT features. With
- this restriction, I see no way to avoid using pointers.
-
- Konrad Hinsen (hac041@zam001.zam.kfa-juelich.de)
-