Hardware/kernel : pipe implementation?

Ok but the Amiga was introduced only much later. I am talking about 1982 up to 1984, and the Amiga only appeared in 1985 I think.

In those days I worked in a small company that developed middleware to run small business applications on Unix systems like the Fortune Systems 16:32, the NCR Tower, the Convergent Technologies Miniframe, etc.

The package included a screen manager (to enter data in screen forms a form at a time to be submitted for processing, like IBM systems did), a database manager (actually more like a keyed filesystem manager with locking), reporting/printing manager, communication with peer systems and local devices, etc.

The whole thing of course was very compact and efficient because in those days you typically had 10 users on a system with 2MB RAM and 40MB disk.

Reply to
Rob
Loading thread data ...

In the early 80s - I was i grade 5 ...

Actually, the system I maintain was developed on vax/vms. It was a port from pdp/11 but with VMS came the use of those mailboxes.

The port to AIX and later on the port to NT did of course not have 'mailboxes' in the VMS sense, so a semaphore/shared mem solution was developed on AIX, where the shared mem was logically divided into mailboxes on NT a mmap()ed file server the same purpose.

Not until a year or so back, I replaced that IPC (that still was hackish) with named pipes, and if each processes has it's own named pipe, we are logically not to far away from the VMS mailboxes again.

Reply to
Björn Lundin

declaimed the following:

Maybe closer to the Amiga message ports...

VMS mailboxes weren't tied to an application process (depending on how it was created); once created they "floated" somewhere in the OS and other processes could connect for both read or write operations {granted, I'm talking about mailboxes created for global access; not private to a process} -- some could exist with no processes connected.

Amiga message ports were one-way -- processes could look up the port by name, and post messages to the port (messages usually included the port address of the posting process for returning status), but the port owner was the reader. The blessing and bane of the Amiga -- a flat address space that was global to all processes; made message ports fast as the message that got posted really only pointed to the address of the data buffer from the process that allocated it (which is why returning the message was desirable, so the deallocation could be done by the same process -- though if one really wanted to, one could deallocate from the recipient side).

ARexx made heavy use of them -- the "address" statement specified essentially a message port name to send data to.

--
	Wulfraed                 Dennis Lee Bieber         AF6VN 
    wlfraed@ix.netcom.com    HTTP://wlfraed.home.netcom.com/
Reply to
Dennis Lee Bieber
[snip]
[snip]

Dave Cutler

Reply to
David James

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.