Art asks:

> Without any side or inferred references ... Weren't the AIF routines
> designed after HP started down the POSIX route?   I seem to remember a
> lot of things coming out of HP that were very much of a *nix
> flavour rather than pure MPE flavor back in the MPE 5 and 6 eras...
...

No, the Architicted Interface Facility (AIF) product was first delivered well before the MPE/iX 4.5 developer release that introduced POSIX APIs. The purpose of AIFs was to provide a high performing, privileged interface to OS routines and data structures that highly skilled application and tools vendors could use to perform their privileged OS interactions in an organized (or architected) manner. The idea was that unlike the "wild west" of MPE V, access to internal routines and data structures should be provided in a manner that helped insulate vendors from internal OS changes that happen from release to release.

By their nature and their design, AIF routines include the smallest possible amount of overhead in order to minimize the penalty for using them rather than directly attempting to access system internals. This means no, or at least very, very little, paramter checking. Because of this, using AIFs requires privileged mode execution. Whenever you enter privileged mode, you are taking responsibility for the integrity of the system. The exact same thing is true of the DEBUG facility. A user with PM capability can go into DEBUG and slam whatever values they want into whatever memory or disk locations they want. But unless you are an extremely competent MPE/iX expert that is usually not a very good idea.

AIFs are an MPE/iX unique invention and have no influence from any other system or API standard.

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *