Jeff King <peff@xxxxxxxx> writes: > I'll stop here for now. It's a fair bit of leg-work digging > these up (though again, I do think there's value in > understanding why the cast was there, even if we know it > isn't _currently_ doing anything). I agree with the value of understanding why each of these casts has become unnecessary, and thanks for demonstrating how a rerolled version should justify its changes with its findings behind each of the unnecessary casts. What do you recommend the next round should look like? Multi patch series, each of which removes one cast with its proposed log message explains how it has become unneeded? A single patch with a gigantic proposed log message that lists the findings for each and all casts that are removed? Somewhere in between, perhaps split along the file boundary, or grouped by the event that made them unneeded (e.g. "cmd_main() used to take non-const but when we made it to take const, all of these casts we remove in this patch became unneeded")? Thanks.
