Re: automake/autoconf/libtool -- convince me
Andreas Metzler <ametzler@downhill.at.eu.org> writes:
> Joey Hess <joeyh@debian.org> wrote:
> [invoking autoconf/automake... at build time instead of shipping the
> results in the diff]
> Quoting autotools-dev/README.Debian.gz
> | And it wastes CPU time in auto-builders, too. Do recall that some
> | architectures are very slow (m68k, for example), and that any wastage
> | of auto-build time is a severe problem for Debian as a whole (it
> | delays packages from moving to Debian testing, for one!).
>
> This might not be a very strong argument but it holds.
No, it doesn't. It's a stupid argument[1]. I can guarantee you that
slower architectures (and I maintain buildds for one of them) will not
notice the noise of autoconf/automake in the big picture.
--
James
[1] And a recurring one; it's the same bogus argument that had policy
mandate brain-dead compile flags for a couple of years. As a
general rule do not believe anyone who starts whining about
auto-builder time. The "time" expense that matters with
auto-builders is always the human cost of processing failures; not
the build time of success builds, and esp. not some small-tiny
portion of the successful build time.
Reply to: