From: Jon Seymour Date: Wed, 03 Aug 2011 15:13:13 GMT Subject: Re: [PATCH v13 5/8] bisect: introduce --no-checkout support into porcelain. Message-ID: In-Reply-To: On Thu, Aug 4, 2011 at 12:09 AM, Christian Couder wrote: > On Wed, Aug 3, 2011 at 3:24 PM, Jon Seymour wrote: >> On Wed, Aug 3, 2011 at 10:46 PM, Jon Seymour wrote: >>> On Wed, Aug 3, 2011 at 3:27 PM, Christian Couder >>> wrote: >>>> On Tuesday 02 August 2011 16:41:13 Jon Seymour wrote: >>>>> On Tue, Aug 2, 2011 at 10:04 PM, Christian Couder >>>>> >>>>> If I was to do this, I'd prefer to change uses of $BISECT_MODE with a >>>>> call to a function bisect_mode() that does the same thing. >>>> >>>> Yeah, I think it would be a good idea to have a bisect_mode() function. >>>> I don't like very much to blindly call some code when we might not need it. >>>> >>> >> Mmmm. >> >> Actually, there is a neater way to do this. >> >> I'll such use the existence of BISECT_HEAD to inform the >> implementation of bisect_mode(). >> >> This avoids the need for a separate .git/BISECT_MODE file. > > Yeah, but then you have to be careful of the fact that BISECT_HEAD > might have not been properly deleted or might have been created by the > user for other purposes. > I have removed $GIT_DIR/BISECT_MODE in v15. If BISECT_HEAD was being used for other purposes, it is going to get deleted anyway, irrespective of whether we have a separate BISECT_MODE file, so I am not sure we need to consider that when deciding when we need a separate BISECT_MODE file. FWIW: bisect_mode() was only going to get called from one place so I just inlined the implementation in that place. (on the call to bisect--helper). jon.