Ralf Wildenhues <Ralf.Wildenhues <at> gmx.de> writes: > > > It looks to me like overall, the shell function changes will make user > > > code even more reliant on correct M4 quotation than before. > > > > Why? > > Because the changes introduce more elaborate M4 constructions overall, > relying more on correct user-side quoting than was needed before. > At least that's my impression, I do not have hard data for this.
I think we've tried to keep the AC_ API unchanged. AS_CASE is actually more tolerant to bad quoting than it used to be. But yes, I concede that bad m4 quoting may bite people where they used to get away with it, due to so many changes in implementation. > > Don't get me wrong, I'm not trying to criticize. The move to shell > function is worth much more than trying to forever support broken > user scripts. However, it should be very clear that the next Autoconf > release quite necessarily is a less stable one for users than the last > one, even if it all it causes is to uncover some more latent issues > in user's configure.ac and .m4 macro files. Long-term, that can only > be a good thing. I'm planning on doing a beta release once we go a week or two without too much churn. Perhaps even calling it Autoconf 2.90 (in preparation for a stable 3.0) rather than 2.64, since we have made so many radical changes in implementation. > > > Anyway, shell functions will have to be mentioned in NEWS. They are. -- Eric Blake
