:wave: If in `Dependencies` I set `aggregating=true` , then “the output should be invalidated by a n...

# ksp

[Wojciech Zięba]

👋 If in Dependencies I set aggregating=true , then “the output should be invalidated by a new source file or a change in any of the existing files”. If so, is there a point of setting originatingFiles in such case, or originatingFiles only makes sense if aggregating=false?

[David Rawson]

IIRC in the case of aggregating = true there is still an optimization available to you if you are careful about originating files. The optimization is for the case where files that are not registered inputs are deleted

[David Rawson]

Screenshot_20240221-071118.png

[David Rawson]

In the diagram above, the black lines from A, B, C to Fascinating.kt represent registration of inputs using originating files for an aggregating output

[Wojciech Zięba]

Thanks David! I’m just watching

your video▾

😄.

The optimization is for the case where files that are not registered inputs are deleted

Gotcha. So still, even with listOf(A,B,C) as originating files, in case of editing D, the reprocessing should happen, is that correct?

[David Rawson]

Editing D.kt could introduce a change that would affect the aggregating output Fascinating.kt that would cause reprocessing. For instance, by adding the annotation of interest.

[David Rawson]

Also, apologies for the poor quality of some of the video content

[Wojciech Zięba]

Not needed, thank you for providing it, it clarified some other things for me. Thank you.

Editing D.kt could introduce a change that would affect the aggregating output Fascinating.kt that would cause reprocessing.

So it’s not any change, but some certain type of change, like adding annotations of interest, do I read it right? E.g. adding other changes, like adding a constructor with println there might not cause reprocessing?

[David Rawson]

I think that from KSP's perspective, any change to D that is not a deletion would invalidate the aggregating output.

[David Rawson]

Invalidation here is about whether changes to the input could potentially harm the correctness of the output. I don't think it's about what the actual changes are (adding println and so on)

[Wojciech Zięba]

Ok, gotcha, it makes sense. Thank you!

[David Rawson]

Other readers of the thread can moderate and point it out if I got it wrong 🤠