My previous misunderstanding was exactly as you mentioned, to group commits on a file to file basis. Now I have understood what a semantic change should be like, with your provided example. You are not asking for much, your corrections help me a lot in my learning journey and I am very grateful for them. Please feel free to point out any more such things in the future. Shiva On 25 August 2023 4:09:09 pm IST, "Michal Prívozník" <mprivozn@redhat.com> wrote:
Again, way too much changes, disperse in semantics for one patch. You've introduced virNetworkObjUpdateModificationImpact(). Perfect! But it should have been one patch. Then you eliminate redundant call to virNetworkObjSetDefTransient()? Splendid, but again - it's a different change and has nothing to do with virNetworkObjSetDefTransient(). You implement new APIs? Sweet, but what do they have to do with the redundant call?
Splitting patches per directory is not the same as "one semantic change per patch". Sometimes it is, e.g. in the series I've posted earlier today:
https://listman.redhat.com/archives/libvir-list/2023-August/241416.html
I know I might be asking too much, but try to put yourself into reviewers shoes. Libvirt's code base is not exactly the smallest and reviewing one change (and trying to think of all implications) is hard enough already. If changes are intertwined into one patch then it's needlessly harder.
Michal