On Tue, 2019-10-01 at 10:12 -0400, Laine Stump wrote:
On 9/30/19 1:35 PM, Andrea Bolognani wrote:
Writing a check that compares the situation before a commit and after it is not as easy as a point-in-time check.
Not all that bad though - just examine the lines that start with +
Fair enough. That wouldn't work outside of git, but I guess it would be an acceptable compromise as all our CI jobs are executed from inside a git checkout anyway.
Instead of spending a non-trival amount of time implementing something like that, I'd rather spend my time dealing with the fallout of a one-time conversion.
Without seeing concrete examples of what actually is "dealing with the fallout" in both cases, I don't want to speculate too much on which would cause more difficulty. I would say that even if we do the conversion all at once, it should be in multiple patches so that if there is some regression caused by the conversion, a git bisect will lead to a multi-hundred line commit instead of a multi-thousand line commit.
Of course when I talk about "big-bang conversion" I mean done in a single series, not a single patch! -- Andrea Bolognani / Red Hat / Virtualization