Gary V. Vaughan wrote: > Hi Jim, > > On 28 Jan 2012, at 16:28, Jim Meyering wrote: >> Gary V. Vaughan wrote: >>> I'm wondering what purpose AC_PREREQ (etc) really >>> serves if you're not using them to encode the versions of the >>> autotools that are >>> required to bootstrap a package in the way expected by the maintainers. >> >> It permits one to build the package unmodified on systems for which >> the latest version of autoconf will never be available. >> This is a big for e.g., libvirt. > > That's a good point. So, will users of libvirt (when using compatible > older autotools > releases) want to be able to run bootstrap?
Yes, though I presume you understand it's developers, not users. If bootstrap itself were to enforce a latest-stable requirement, it would cause grief for any developer not using the equivalent of Fedora rawhide or Debian's unstable -- and sometimes even those lack the latest stable version of some build tool, though not for long. > If so, then my implementation in saner bootstrap is correct, and > keeping the paragraph > in README-release we're discussing is correct too... but I need to > revise my thinking > about AC_PREREQ and friends a little, and not misuse them to name latest > stable > releases. And also make a point of testing old versions of autotools I wouldn't call it "misuse". It's a compromise to accommodate developers who build from cloned sources. If no one has complained, then it's probably ok (and IMHO recommended) to require the latest versions. > to reaffirm the > correctnesss of AC_PREREQ (etc), which is something I stopped doing a > while ago as it > consumes a lot of time, for (what I then thought) was little tangible gain.
