Re: cross-install-strip summary

[Tom Tromey]

>>>>> "adl" == Alexandre Duret-Lutz <address@hidden> writes:

adl> I have tested install-strip for cross-compilation with CVS
adl> Automake, doing both an in-place and a VPATH build (with relative
adl> paths, I have not tested absolute paths).  It works here, great
adl> :)

Good.

adl>        - $INSTALL_STRIP_PROGRAM assumes install-sh, not install.sh
adl>          (easily fixed)

install.sh has been deprecated for years now.  We warn about it.
Supporting it here is unimportant.

adl>        - $_am_dirpart can fooled when $install_sh is defined
adl>          using $am_missing_run

When does this happen?

adl>        - $STRIP has to be set *before* AM_INIT_AUTOMAKE is called
adl>          (either by the user or in a configure check); I'm not
adl>          really happy with that

Me neither.  What if we make install-strip unconditionally use
install-sh?  Do we really care about the small performance impact on a
rarely-used target?

adl>    * install-strip when *not* cross-compiling:
adl>        - $INSTALL_STRIP_PROGRAM uses sub make evaluation, does it
adl>          really matters?

It doesn't matter.

Tom