Re: Initial start of a PCD template

[yyyyyyyyyy@xxxxxxx]

In a message dated 2003-08-07 7:10:08 PM Eastern Daylight Time, Larry.Dwyer@xxxxxx writes:

> As Andrew (correctly) pointed out in the mail "Defect in XBD 
> 3.386_System_Reboot" et al, these things which are "implementation defined" 
> cannot be used by a portable application so they are irrelevant, or more 
> properly put, they should be "unspecified".  This is the point I was trying 
> to make (rather poorly).
>
>          Cheers,
>          Larry


The way I see it, implementation-defined implies that it may not be possible to use the same code to perform the same function on all conforming implementations, but it is reasonable for an application writer to expect some degree of this functionality being available. For the easier cases this may be done intrinsically via the conf() interfaces, or getconf in scripts; the not so easy via #ifdef after inspecting the documentation. The cases brought up here are representative of the latter category, and seem to be more indicative of placeholders for things that were identified to have a concrete logical model but it was not desired to burden implementors from taking advantage of hardware advances or other technology changes, not that they are intrinsically unspecifiable or of no tangible use to members of the standards' target audiences.

Cheers,
Mark