Hi, On Wed, 23 Jan 2008, Kirill wrote:
> On Jan 23, 7:42 am, Johannes Schindelin wrote: > > > Basically because I am way more familiar with standard C libraries. > > Is there a performance hit, or any other problem with mixing [standard > > C libraries and Win32 API]? > > I'm not aware of any performance or any other issues, except: > - a stronger dependency on a particular implementation of C run-time > libraries; > - a requirement to remember two sets of API. Given that I work much more on other platforms than Windows, and the the standard C library is available even on Windows (that's why it is called "standard", even if Microsoft succeeded in introducing subtle differences), it would not be wise for me to start leaning to the Windows API and be constantly confused. And given the fact that most contributors that I am aware of (which amounts to Brian and me, mostly) are in this situation, I do not think it would be wise to change git-cheetah in that respect now; if most patches were done by a person intrinsically familiar with the Windows API, I might have agreed with you. > It's just a personal preference of keeping the number of commits > "reasonable", even though Linus seems to prefer short commits. It's just > three short commits in two days look like an overkill. Has nothing to do with overkill. It helps improve bisectability (IOW bug-hunting). So even if you are doing the patch, I prefer one-issue-per-commit. > P.S. Johannes and Johannes, please, accept my apologies if you find the > following question offensive. Do you mind if one uses your signatures > [instead of your names] when greeting you? I have no problem with you calling me "Dscho". And I usually make a point of calling Hannes "Hannes", just to make obvious whom I mean: on this list, there is only one "Hannes". (And if a new "Hannes" should come along, well, too bad, the name is already taken ;-) Ciao, Dscho
