Re: [PATCH 22/38] arm: dts: qcom: ipq8064: Break long property array data

[Waqar Hameed]

On Fri, Oct 16, 2026 at 16:48 +0200 Krzysztof Kozlowski <krzk@xxxxxxxxxx> wrote:

> On 16/10/2026 16:15, Waqar Hameed wrote:
>> Yeah, as I responded in patch 19, this is hard and needs some
>> discussions. Having the braces on separate lines like this is one way to
>> do it, and in fact that's how `black` for Python does it. `rustfmt` also
>> chooses this but first tries to fit it on one line with a newline after
>> the assignment:
>> 
>> 		interrupts =
>> 			<GIC_PPI 10 (GIC_CPU_MASK_SIMPLE(2) | IRQ_TYPE_LEVEL_HIGH)>;
>> 
>> If it still doesn't fit 100 chars (like here), it then formats long
>> arrays as
>> 
>> 		interrupts = <
>> 			GIC_PPI 10 (GIC_CPU_MASK_SIMPLE(2) | IRQ_TYPE_LEVEL_HIGH)
>> 		>;
>> 
>> This still doesn't fit 100 chars. Maybe we could then wrap the
>> individual items like:
>> 
>> 		interrupts = <
>> 			GIC_PPI 10
>> 			(GIC_CPU_MASK_SIMPLE(2) | IRQ_TYPE_LEVEL_HIGH)
>> 		>;
>> 
>> I think that's a reasonable algorithm to have in dtfmt for wrapping
>> arrays, or?
>
> The code should be:
> interrupts = <GIC_PPI 10 (GIC_CPU_MASK_SIMPLE(2) | IRQ_TYPE_LEVEL_HIGH)
> it fits within reasonable limits. Why would you wrap it?

Ah sorry, I can see how this is confusing, but this was just a response
to Konrad's comment "... For lines that overflow, I think we may need
some brainstorming from the wider audience". This was my "brainstorming"
about line wrapping in general. What do you think Krzysztof about this
approach?

We are all in agreement that the line in _this patch_ should only be on
one line, without any wrapping (since it's shorter than 100 chars)!