From: Keith M. <kei...@to...> - 2005-09-26 16:40:02
|
Angus Leeming wrote: > Here's my version of your macro. It differs only in the test that leads > to ac_pwd_w being initialised which is now your improved solution from > elsewhere in this thread. > > Is this 'improvement' correct, or was your original the definitive > solution? I'd like to think about that a bit. In the context of the macro, the extended test isn't necessary, and adds an extra fork in `configure'. My initial thought is that my original should be considered definitive; we don't need to ensure that `pwd -W' returns a Win32 path, within the macro's context -- all that is required is an absolute path, which has the appropriate semantics for the host on which `configure' is run. Note that this macro won't give the desired result, if cross-compiling to a `mingw32' host, from a foreign build platform, but the extended test isn't sufficient either, in this situation. Best regards, Keith. |