Bruno Haible <br...@clisp.org> wrote: ... >> It looks to me like the change below is equivalent to yours, > > Ah, I see now what you mean. Fine with me. > > I wouldn't have chosen this solution because it tears apart the > determination of the ac_cv_func_working_mktime into two parts, > one before the AC_CACHE_CHECK and one inside it (making the logic > more complex)
I don't see the tearing, or any detrimental effect on the logic. Either way, there is a test of $APPLE_UNIVERSAL_BUILD before the cache check. The only difference is that I prefer to move the ac_cv_func_working_mktime=no assignment "up", so it's right next to that test. > and because the configure output will show > > checking for working mktime... (cached) no > > already at the first invocation of the macro. It might be worth your while to make the macro emit an explanatory diagnostic. The point is to have the test of $APPLE_UNIVERSAL_BUILD and the corresponding ac_cv_func_working_mktime=no assignment in close proximity. With the latter patch, the lines are essentially adjacent. The other way, they're almost 200 lines apart. >> Then the nesting/logic of the majority of lines in >> that file doesn't change. > > I find it normal to reindent even large chunks of code, when needed Same here. The key is "when needed".
