Allan McRae wrote:
FC wrote:
From ae3dd41af891f059b0aa1cc888eed873684b5580 Mon Sep 17 00:00:00 2001 From: Francois Charette <francois@archlinux.org> <francois@archlinux.org> Date: Tue, 18 Aug 2009 22:10:56 +0200 Subject: [PATCH] makepkg -R: allow unpacking when pkgdir is empty
Also fixed some typos :)
When you "additional" part of your patch is as large as the patch itself, just make two patches. I have not reviewed the typo fixes.
Fair enough. Should I resend as two separate patches?
create_package() { if [ ! -d "$pkgdir" ]; then - error "$(gettext "Missing pkg/ directory.")" - plain "$(gettext "Aborting...")" - exit 1 # $E_MISSING_PKGDIR + if [ "$REPKG" -eq 1 -a -f $PKGDEST/${pkg}-${pkgver}-${pkgrel}-${CARCH}${PKGEXT} ]; then + mkdir "$pkgdir" && cd "$pkgdir" + if ! bsdtar -xf $PKGDEST/${pkg}-${pkgver}-${pkgrel}-${CARCH}${PKGEXT} ; then
Please keep test style consistent with the rest of the code base.
Not sure what you mean...
+ error "$(gettext "Failed to extract package for repackaging")" + exit 1 # TODO: error code + fi + cd ..
Do not "cd .." as $startdir and $pkgdir are not always related like that. Especially with split packages.
ok
+ else + error "$(gettext "Missing pkg/ directory.")" + plain "$(gettext "Aborting...")" + exit 1 # $E_MISSING_PKGDIR + fi fi
cd "$pkgdir"
General comment: Have you tested this with PKGBUILDs that define a package() function (including split packages)? These PKGBUILDs run the package() function when using the -R option. Without actually testing, I think this will fail.
Right, this will certainly not work with split packages... Before I spend more time on this, do you think this is worth implementing? I thought it coult be useful in situations where you build a pkg (with makepkg -c) that takes a long time compiling, and then realize you forgot one dependency, for instance. Then makepkg -R would work even if the pkgdir is empty. Of course it is also no big deal to unpack manually first... F