[pacman-dev] [PATCH] makepkg: allow specifying subsets of packages to build

[Allan McRae]

Dan McGee wrote:
> On Tue, Oct 20, 2009 at 11:01 PM, Allan McRae <allan at archlinux.org> wrote:
>   
>> This allows makepkg to only build a specified subset of packages
>> from a split PKGBUILD.  This is very useful in combination with the
>> -R flag or when bumping the pkgrel of a single package.
>>
>> Fixes FS#15956.
>>
>> Signed-off-by: Allan McRae <allan at archlinux.org>
>> ---
>>
>> As always, my changes to the documenation could be improved.  Suggestions welcome.
>>     
>
> Would you mind splitting the ARGLIST patch into a separate one? It
> seems pretty independent and might be good to explain a bit more by
> itself in a commit message. (Easiest way to split a patch is probably
> git-reset HEAD^, then git add -i, ...)
>   

No problem.

>   
>>  doc/makepkg.8.txt     |    5 +++++
>>  scripts/makepkg.sh.in |   26 +++++++++++++++++++++-----
>>  2 files changed, 26 insertions(+), 5 deletions(-)
>>
>> diff --git a/doc/makepkg.8.txt b/doc/makepkg.8.txt
>> index ec02875..8c68143 100644
>> --- a/doc/makepkg.8.txt
>> +++ b/doc/makepkg.8.txt
>> @@ -148,6 +148,11 @@ Options
>>        remote builder, or a tarball upload. Because integrity checks are verified,
>>        all source files of the package need to be present or downloadable.
>>
>> +*\--pkg <`list`>*::
>> +       Builds only the listed packages. Useful in combination with package
>> +       splitting.  The use quotes is necessary when specifying multiple
>> +       packages. e.g. `--pkg "pkg1 pkg3"`
>> +
>>     
> If I wasn't reading this with split packages in mind, I might think
> this is some sort of yaourt automatic system where it goes an finds
> the package I want. "Builds only the designated packages in a split
> package build script" or something? I don't know. Now of course I just
> read the second sentence and see that you cleared it up; it might be
> worth trying to get the package splitting actually in the first
> sentence for stupid people like me that only read the first sentence.
> And "The use of quotes" might be what you were looking for.
>   



>   
>>  *\--noconfirm*::
>>        (Passed to pacman) Prevent pacman from waiting for user input before
>>        proceeding with operations.
>> diff --git a/scripts/makepkg.sh.in b/scripts/makepkg.sh.in
>> index 7e10eef..c3672b1 100644
>> --- a/scripts/makepkg.sh.in
>> +++ b/scripts/makepkg.sh.in
>> @@ -72,6 +72,7 @@ IGNOREARCH=0
>>  HOLDVER=0
>>  PKGFUNC=0
>>  SPLITPKG=0
>> +PKGLIST=""
>>  COLORMSG=0
>>
>>  # Forces the pkgver of the current PKGBUILD. Used by the fakeroot call
>> @@ -1243,6 +1244,15 @@ check_sanity() {
>>                done
>>        fi
>>
>> +       if [ -n "$PKGLIST" ]; then
>>     
> This would only check the first item in $PKGLIST, is that ok?
>   

I think that is fine.

>> +               for pkg in ${PKGLIST[@]}; do
>> +                       if ! in_array $pkg ${pkgname[@]}; then
>> +                               error "$(gettext "requested package %s is not provided in %s")" "$pkg" "$BUILDSCRIPT"
>> +                               return 1
>> +                       fi
>> +               done
>> +       fi
>> +
>>        return 0
>>  }
>>
>> @@ -1466,6 +1476,7 @@ usage() {
>>        echo "$(gettext "  --asroot         Allow makepkg to run as root user")"
>>        printf "$(gettext "  --config <file>  Use an alternate config file (instead of '%s')")\n" "$confdir/makepkg.conf"
>>        echo "$(gettext "  --holdver        Prevent automatic version bumping for development PKGBUILDs")"
>> +       echo "$(gettext "  --pkg <list>     Only build listed packages from a split package")"
>>     
> So for the man page doc, this sentence is maybe succinct enough to be
> the lead sentence, then follow it with more detail.
>
>   
>>        echo "$(gettext "  --skipinteg      Do not fail when integrity checks are missing")"
>>        echo "$(gettext "  --source         Generate a source-only tarball without downloaded sources")"
>>        echo
>> @@ -1496,13 +1507,13 @@ if [ ! $(type -t gettext) ]; then
>>        }
>>  fi
>>
>> -ARGLIST=$@
>> +ARGLIST=("$@")
>>
>>  # Parse Command Line Options.
>>  OPT_SHORT="AcCdefFghiLmop:rRsV"
>>  OPT_LONG="allsource,asroot,ignorearch,clean,cleancache,nodeps"
>>  OPT_LONG="$OPT_LONG,noextract,force,forcever:,geninteg,help,holdver"
>> -OPT_LONG="$OPT_LONG,install,log,nocolor,nobuild,rmdeps,repackage,skipinteg"
>> +OPT_LONG="$OPT_LONG,install,log,nocolor,nobuild,pkg:,rmdeps,repackage,skipinteg"
>>  OPT_LONG="$OPT_LONG,source,syncdeps,version,config:"
>>  # Pacman Options
>>  OPT_LONG="$OPT_LONG,noconfirm,noprogressbar"
>> @@ -1540,6 +1551,7 @@ while true; do
>>                -m|--nocolor)     USE_COLOR='n' ;;
>>                -o|--nobuild)     NOBUILD=1 ;;
>>                -p)               shift; BUILDFILE=$1 ;;
>> +               --pkg)            shift; PKGLIST=$1 ;;
>>                -r|--rmdeps)      RMDEPS=1 ;;
>>                -R|--repackage)   REPKG=1 ;;
>>                --skipinteg)      SKIPINTEG=1 ;;
>> @@ -1722,6 +1734,11 @@ fi
>>
>>  pkgbase=${pkgbase:-${pkgname[0]}}
>>
>> +if [ -n "$PKGLIST" ]; then
>> +       unset pkgname
>> +       pkgname="${PKGLIST[@]}"
>> +fi
>> +
>>  if [ "$SPLITPKG" -eq 0 ]; then
>>        if [ \( -f "$PKGDEST/${pkgname}-${pkgver}-${pkgrel}-${CARCH}${PKGEXT}" \
>>             -o -f "$PKGDEST/${pkgname}-${pkgver}-${pkgrel}-any${PKGEXT}" \) \
>> @@ -1914,11 +1931,10 @@ else
>>                fi
>>
>>                msg "$(gettext "Entering fakeroot environment...")"
>> -
>>                if [ -n "$newpkgver" ]; then
>> -                       fakeroot -- $0 --forcever $newpkgver -F $ARGLIST || exit $?
>> +                       fakeroot -- $0 --forcever $newpkgver -F "${ARGLIST[@]}" || exit $?
>>                else
>> -                       fakeroot -- $0 -F $ARGLIST || exit $?
>> +                       fakeroot -- $0 -F "${ARGLIST[@]}" || exit $?
>>                fi
>>        fi
>>  fi
>> --
>> 1.6.5.1
>>     
>
> Without testing, I'm not sure how the flipflopping between treating
> $PKGLIST like a string and an array works, but it seems scary. Am I
> missing something obvious about the way bash does this?
>   

The short answer is that it works...  although I am sure someone could 
find and edge case where it does not.  I will adjust it to use array 
syntax always.

Allan