On 07/09/14 21:37, David Aguilar wrote: > On Sun, Sep 07, 2014 at 12:01:00PM +0100, Ramsay Jones wrote: >> On 07/09/14 11:35, René Scharfe wrote: >>> Am 07.09.2014 um 11:36 schrieb David Aguilar: >>>> Add dependent headers so that including a header does not >>>> require including additional headers. >>>> >>>> This makes it so that "gcc -c $header" succeeds for each header. >>> >>>> diff --git a/cache.h b/cache.h >>>> index 4d5b76c..8b827d7 100644 >>>> --- a/cache.h >>>> +++ b/cache.h >>>> @@ -1,7 +1,6 @@ >>>> #ifndef CACHE_H >>>> #define CACHE_H >>>> >>>> -#include "git-compat-util.h" >>>> #include "strbuf.h" >>>> #include "hashmap.h" >>>> #include "advice.h" >>> >>> Oh, that's a new change and worth mentioning in the commit message. >> >> Hmm, does this not break git? Unless you also change each '.c' file which >> includes cache.h to also include git-compat-util.h first, then I suspect >> (if nothing else) file I/O may be broken. (see _FILE_OFFSET_BITS). >> >> Also, see Documentation/CodingGuidelines (lines 331-333). > > This one should be okay because hashmap.h includes git-compat-util.h. It is important that each compilation unit include 'git-compat-util.h' at the start. From the CodingGuidelines we see that 'cache.h' and 'builtin.h' are *blessed* headers which can be used instead. (I have always thought that this was a mistake; it would be much easier to just stick to using 'git-compat-util.h' at the top of each '.c' file ;-) The fact that hashmap.h happens to include git-compat-util.h is not at all relevant here. > > Jonathan will be re-rolling so I'll definitely read and test the > patches when they're ready. They'll probably be a more focused, > surgical change then this version. OK. ATB, Ramsay Jones -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html