Ralf Wildenhues wrote, quoting me: >>> >>> That is not strictly true. It'd undocumented (and as such you >>> shouldn't rely on it), but you can >>> AC_REQUIRE([FOO], [FOO([arg])]) >> >> ... >> >> This suggested usage isn't simply undocumented; it actually conflicts >> with explicitly documented behaviour. Even if is is possible, (and I've >> no doubt that it is), do you really consider it wise to even suggest it, >> particularly since you also caution that it should not be relied on? > > I'm not suggesting it. The OP should definitely not use it. Sorry > for mentioning it. I merely wanted to avoid somebody "finding that > information" in the Autoconf source code, and using it. So I warned > against it that it's not to be relied on. Ah, sorry. I misunderstood your intention in mentioning it, (and so others may have too). I read it as a suggestion that it was something that might be considered as useful, albeit potentially risky. > Enough warning signs now? Yep. > To be even more explicit: we would like that the users of Autoconf > *only* rely on interfaces documented in autoconf.info. No arguments on that; I agree 100%. Regards, Keith. _______________________________________________ Autoconf mailing list Autoconf@xxxxxxx http://lists.gnu.org/mailman/listinfo/autoconf
