On Sun, Nov 11, 2018 at 11:16:50AM +0100, Torsten Bögershausen wrote: > > I like the overall direction. I feel a little funny doing this step now, > > and not as part of a series to convert individual variables. But I > > cannot offhand think of any reason that it would behave badly even if > > the other part does not materialize > > > > Hej all, > There may be some background information missing: > - I did a 2-patch series based on this commit in pu: > commit 37c59c3e8fac8bae7ccc5baa148b0e9bae0c8d65 > Author: Junio C Hamano <gitster@xxxxxxxxx> > Date: Sat Oct 27 16:42:25 2018 +0900 > > treewide: apply cocci patch > > (that patch was never send out, see below) > .... > The week later, I tried to apply it on pu, but that was nearly hopeless, > as too much things had changed on pu. > I had the chance to compile & test it, but decided to take "part2" before > "part1", so to say: > Fix all the printing, and wait for the master branch to settle, > and then do the "unsigned long" -> size_t conversion. > That will probably happen after 2.20. Ah, OK. I am fine with that approach. My thinking was that we'd see individual functions and their callers converted, which is another way to do it incrementally. But sometimes that ends up cascading and you end up having to change quite a bit of the callstack anyway. -Peff
