
|
[openbeosstorage]
||
[Date Prev]
[04-2003 Date Index]
[Date Next]
||
[Thread Prev]
[04-2003 Thread Index]
[Thread Next]
[openbeosstorage] Re: DiskDevice API v2.1
- From: "Ingo Weinhold" <bonefish@xxxxxxxxxxxxxxx>
- To: openbeosstorage@xxxxxxxxxxxxx
- Date: Sat, 05 Apr 2003 18:31:43 +0200 CEST
"Axel D=F6rfler" <axeld@xxxxxxxxxxxxxxxx> wrote:
>
> Tyler Dauwalder <tyler@xxxxxxxxxxxxx> wrote:
> > > Yes, the caller would have to make sure, that the parameters are
> > > valid.
> > Doesn't it seem like it would be clearer and less error prone just
> > to
> > have a specific function per operation=3D3F Yes, that means more
> > functions,
> > but IMHO that's a reasonable tradeoff for better typechecking and
> > greater ease of implementation.
>
> It depends on where this generic function would be. From userland, we
> could have 20 functions that will all create the parameter set for
> this
> one generic function, which is a syscall.
We could also have 20 syscalls, which are mapped to one FS API
function. That would allow for easy kernel side parameter checking.
> The underlying FS could also have this one generic function. Having
> only one generic function (or a small set of functions) would make
> the
> FS API easier.
That's what I thought too.
CU, Ingo
|

|