> On Mar 2, 2015, at 1:18 PM, James Peach <jpe...@apache.org> wrote: > >> >> On Feb 28, 2015, at 8:57 PM, Leif Hedstrom <zw...@apache.org> wrote: >> >> >>> On Feb 28, 2015, at 9:35 PM, Phil Sorber <sor...@apache.org> wrote: >>> >>> On Wed Feb 25 2015 at 9:15:38 AM James Peach <jpe...@apache.org> wrote: >>> >>>> >>>>> On Feb 24, 2015, at 1:59 PM, Leif Hedstrom <zw...@apache.org> wrote: >>>>> >>>>> Hi, >>>>> >>>>> clang-format has finally gotten to the point where we can get it to >>>> format our code similar, but not quite identical, to what we have today. >>>> Doing all formatting programmatically has several benefits: >>>>> >>>>> It’s no longer up to subjective or personal preferences, we’ll learn to >>>> live and love the clang-format coding style. >>>>> It can be automated. >>>>> It can also be used as a tool for people who want to work / see code in >>>> a different style, but commit in our standard style. >>>>> >>>>> >>>>> I have updated the .clang-format files that is in our Git master, we >>>> might need to do a few more tweaks, but it’s getting pretty close. It does >>>> require a very recent version of clang-format, the one I used is >>>>> >>>>> clang-format version 3.6.0 (tags/google/testing/2015-01-13) >>>> >>>> Does this completely destroy code history? >>>> >>>> J >>> >>> >>> The simple answer to this is: "No, we still use git." >>> >>> But what I think you are really asking is, "Does `git blame` become less >>> useful?" >>> >>> While it's true that a simple `git blame` will show lots of format changes >>> instead of what you may deem more useful, I would argue there are better >>> ways to find what you are looking for anyway: >>> http://jfire.io/blog/2012/03/07/code-archaeology-with-git/ >> >> >> Yeah, that’s a good article. Even a simple -w -M fixes most of the pain. On >> a clang-format’ed file, e.g. >> >> heimdall (21:56) 300/0 $ git blame proxy/http/HttpSM.cc | grep 'Leif' | wc -l >> 1465 >> heimdall (21:56) 301/0 $ git blame -w -M proxy/http/HttpSM.cc | grep 'Leif' >> | wc -l >> 680 > > but top of tree is > > jpeach$ git blame -w -M proxy/http/HttpSM.cc | grep 'Leif' | wc -l > 382 > > So it seems like you will still lose some history, but when I manually > compared the blame before and after it didn't seem too bad
Yeah, -w -M doesn’t solve it entirely, but it’s a good start for sure. There are a whole slew of other things to help track down those cases where history is hidden (not lost, history is never lost, unless someone does a Geffon on the git repo :). Speaking of which, how do I make “-w -M” the default for git-blame? Do I have to make an alias for it? Cheers, — Leif
