On Fri, Aug 19, 2022 at 05:17:17PM +0300, Krzysztof Kozlowski wrote:
On 19/08/2022 17:14, Martin Povišer wrote:
Since it was brought up last time but I didn’t respond: the nonalphabetical order is as the chips were introduced (and matches other schemas).
Sure, just keep that order for future compatibles as well - so always put them according to verifiable time of market introduction...
This is very poor reason, instead of alphabetical order. Even worse reason is repeating wrong pattern just because someone else did it.
Best regards, Krzysztof
I don’t see it nearly as clear-cut. Adding to the end seems pretty foolproof too, but OK, next submission will have it alphabet. ordered.
The concept is the same everywhere, be it Kconfig, Makefile or other lists. If everyone adds at the end, you increase the chances of conflicts. Having alphabetical order usually means simultaneous edits will happen in different places.
The difference for those cases is there is 0 control of when things are added with the source being all independent (different companies). For these, it's all one platform family and there's limits as to when one source can produce new entries.
I'd kind of like to know timeline order, but alphabetical is the only thing we can ever check easily and possibly automate (hint).
Rob