Re: [PATCH 1/2] ad1848: separate header for registers definition

[Rene Herman <rene.herman@keyaccess.nl>]

On 15-06-08 12:41, Krzysztof Helt wrote:

> This patch is a start combining ad1848_lib and cs4231_lib
> functionality. My approach is to make the cs4231_lib module
> use the ad1848_lib module. This should allow removing
> most of identical functions from the cs4231_lib and use
> the functions from the ad1848_lib directly.
> 
> If this approach is not a good idea, please drop the patch.
> 
> My current plan is to reuse the ad1848 register constants
> in the cs423x code and remove duplicated cs4231 constants.
> This should make the ad1848_lib and cs4231_lib easier to
> compare for differences.
> 
> The next step is to use the snd_cs4231 structure instead of 
> the snd_ad1848 structure inside the ad1848_lib. This will make
> functions from the ad1848_lib taking the same arguments
> as the functions in the cs4231_lib.
> 
> The last step would be to export needed functions from the
> ad1848_lib and use them directly in the cs4231_lib.
> 
> This should reduce the size of the cs4231_lib while preserving
> smaller the ad1848_lib for driver which uses only ad1848
> functionality.

I believe it would be going a bit overboard to actually make these two 
dependent libs. Unification would seem a good plan probably but as far 
as I'm concerned one lib to drive all these chips should do really.

cs4321_lib isn't all that much bigger. The seperation does make sense at 
the source level, but I'd personally go for a single library module (*) 
built from multiple source files. These chips aren't used by embedded 
these days and even if they were the sizes we are talking about are 
laughable in today's terms.

At a later stage, cs4236_lib could be folded in as well...

(*) perhaps under sound/drivers/wss, or sound/isa/lib/wss (or if you 
don't care for the wss name, something else; the location is more the 
point).

Rene.