[Bug build/18512] make install failure with overridden prefix

["carlos at redhat dot com" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=18512

--- Comment #4 from Carlos O'Donell <carlos at redhat dot com> ---
(In reply to Martin Sebor from comment #3)
> In section 7.2.5 Variables for Installation Directories, the GNU Coding
> Standard describes the prefix fix variable as the mechanism expected to be
> used to override the standard installation directory.  If GLIBC doesn't
> supported this use case (which I would find understandable), I propose
> mentioning it in GLIBC documentation (i.e., in the INSTALL file and online).
> In fact, I already have a patch for INSTALL to that effect but have been
> waiting for feedback on this issue before submitting it.  With the patch
> accepted I think the issue can be resolved.

You are correct, it is indeed in the GNU Coding Standards, and while we strive
to adhere to the standard where possible, in this particular case we don't have
the infrastructure to support installing to an alternate prefix without a
recompilation. To do so would require a data-driven approach to determe the
prefix at runtime, and while such an approach would actually be interesting for
testing, nobody has implemented it yet (chroot, containers, or vm's are used
instead).

I would support a patch to INSTALL to correct this. Though I would like it even
more if `make install prefix=/foo` failed if /foo didn't match the configure's
prefix argument.

-- 
You are receiving this mail because:
You are on the CC list for the bug.