the log message for the commit that introduced this change indicates that it
was necessary for some reason - i think it would be hasty to revert this until
we know what the trade-off would be
"- We mount the temporary directory containing the extracted source package
files read-only, to be sure that makepkg doesn't modify the PKGBUILD. This is
necessary because --holdver only disables pkgver() if it's a VCS package."
right now, i think the only packages that are broken by this are the kernels -
there is a single command that requires a writable /startdir, and it does not
appear to have any value whatsoever - all that it does is write the 'pkgbase'
into the .install script - 'pkgbase' will never change, so that could and
should be hard-coded into the .install script - i suggest this change to the
kernels PKGBUILDs instead
--- a/linux.install
+++ b/linux.install
@@ -7,8 +7,6 @@
}
post_remove() {
- rm -f boot/initramfs-%PKGBASE%.img
- rm -f boot/initramfs-%PKGBASE%-fallback.img
+ rm -f boot/initramfs-linux-libre.img
+ rm -f boot/initramfs-linux-libre-fallback.img
}
-
-# vim:set ft=sh ts=8 sts=2 sw=2 et:
--- a/PKGBUILD
+++ b/PKGBUILD
@@ -10,7 +10,7 @@
# Based on linux package
pkgbase=linux-libre # Build stock kernel
-#pkgbase=linux-libre-custom # Build kernel with a different name
+#pkgbase=linux-libre-custom # Build kernel with a different name (ASSERT:
hard-coded into linux.install)
_srcbasever=5.1-gnu
_srcver=5.1.3-gnu
@@ -225,10 +225,6 @@
"
fi
- # hack to allow specifying an initially nonexisting install file
- sed "$subst" "$startdir/$install" > "$startdir/$install.pkg"
- true && install=$install.pkg
-
# fill in mkinitcpio preset and pacman hooks
sed "$subst" ../linux.preset | install -Dm644 /dev/stdin \
"$pkgdir/etc/mkinitcpio.d/$pkgbase.preset"
_______________________________________________
Dev mailing list
[email protected]
https://lists.parabola.nu/mailman/listinfo/dev
