[interfacekit] Re: :) Again - BBitmap

On Fri, 25 Jul 2003 15:46:05 +0200 CEST "Axel Dörfler" <axeld@pinc-
software.de> wrote:
> "burton666@xxxxxxxxx" <burton666@xxxxxxxxx> wrote:
> > ... client's side. But currently it's not available and won't be 
> > for 
> > some time, 
> > because before releasing a new API of that importance, we want to 
> > clean 
> > things up. Because once it's done, it's done. We have to stay 
> > compatible 
> > with it for a long time. "
> > 
> > In my opinion, we should stay as much close as possible at R5, at 
> > least in 
> > our first release.
> > We chose the R5 design so we wouldn't have wasted time in design 
> > work, 
> > right ? 
> 
> Our idea was not to create a new API, but to stretch the existing 
> BView 
> API to suit our needs - IMO, that's a much much nicer way.

Yep. Though I have to agree, that I wouldn't do too much fine tuning 
till R1.

> > An example of the desing time we could "waste" is represented by 
> > the 
> > private storage api Ingo has implemented. Just look back at the 
> > time/
> > work
> > Ingo has put into them, since we weren't going to follow R5 design 
> > (I 
> > can't
> > say if it's been a good or bad decision, that's just a fact).

Yes, designing a non-trivial API properly takes a lot of time.

The decision not to with the R5 design regarding the Disk Device 
(formerly known as DeviceMap) API was made, because there simply was 
nothing that I would call a design. The DeviceMap API was simply a 
wrapper for the DriveSetup add-ons, whose interface wasn't even (still 
isn't) completely known to us. Moreover there was the requirement, that 
the functionality to recognize partitions should also be available in 
the kernel/boot loader. So, our first approach was to only change a few 
things of the API, but implement the guts quite different (small set of 
functionality in the kernel, and a lot in the registrar). Unfortunately 
the design had some flaws (not so serious to make the thing not work, 
but at least not nice) and Tyler came up with an excellent new idea for 
the API, so we decided to bite the bullet, drop the old stuff, and 
restart again.

It would have been best to do that already in the beginning, but you 
know, it doesn't always work out as one would like it to.

> Yes, we gave him a hard time :-)

I can tell you! :-)

> But I am sure it was worth it :))

Definitely, the API and its implementation is much nicer than in R5 
(and certainly also Dano). Another thing the YellowTab guys will be 
envious of. ;-)

CU, Ingo

Other related posts: