[arch-projects] [devtools] [PATCH] add ccache support

[Dominik Heidler]

On Tue, Jul 30, 2013 at 4:08 PM, Dave Reisner <d at falconindy.com> wrote:

> On Tue, Jul 30, 2013 at 03:55:26PM +0200, Dominik Heidler wrote:
> > Signed-off-by: Dominik Heidler <dheidler at gmail.com>
> > ---
> >  makechrootpkg.in | 13 ++++++++++++-
> >  1 file changed, 12 insertions(+), 1 deletion(-)
> >
> > diff --git a/makechrootpkg.in b/makechrootpkg.in
> > index d7d3ecf..d4c6abe 100644
> > --- a/makechrootpkg.in
> > +++ b/makechrootpkg.in
> > @@ -21,6 +21,8 @@ run_namcap=false
> >  temp_chroot=false
> >  chrootdir=
> >  passeddir=
> > +ccachedir=
> > +passedccdir=
> >  declare -a install_pkgs
> >  declare -i ret=0
> >
> > @@ -58,10 +60,11 @@ usage() {
> >       echo "           Default: $copy"
> >       echo '-n         Run namcap on the package'
> >       echo '-T         Build in a temporary directory'
> > +     echo '-C <dir>   The ccache directory to use'
> >       exit 1
> >  }
> >
> > -while getopts 'hcur:I:l:nT' arg; do
> > +while getopts 'hcur:I:l:C:nT' arg; do
> >       case "$arg" in
> >               h) usage ;;
> >               c) clean_first=true ;;
> > @@ -71,6 +74,7 @@ while getopts 'hcur:I:l:nT' arg; do
> >               l) copy="$OPTARG" ;;
> >               n) run_namcap=true; makepkg_args="$makepkg_args -i" ;;
> >               T) temp_chroot=true; copy+="-$$" ;;
> > +             C) passedccdir="$OPTARG" ;;
> >               *) makepkg_args="$makepkg_args -$arg $OPTARG" ;;
> >       esac
> >  done
> > @@ -84,6 +88,12 @@ chrootdir=$(readlink -e "$passeddir")
> >  [[ ! -d $chrootdir ]] && die "No chroot dir defined, or invalid path
> '$passeddir'"
> >  [[ ! -d $chrootdir/root ]] && die "Missing chroot dir root directory.
> Try using: mkarchroot $chrootdir/root base-devel"
> >
> > +# Canonicalize ccachedir, getting rid of trailing /
>
> Ok, but why is this necessary? Based on the comment, I'd say you just
> want to make the assignment dir=${dir%/} and readlink is overkill. But,
> I still don't understand why this is needed.
>
> Ranty/semi-offtopic: Comments should explain why, not what. If the
> "what" needs explaining, then your code is probably too complicated. I
> realize bash is an odd case and sometimes the syntax itself needs
> clarification, but that's clearly not the case here.
>
> > +if [[ "$passedccdir" != "" ]]; then
>
> if [[ -z $passedccdir ]]; then ...
>
> > +     ccachedir=$(readlink -e "$passedccdir")
> > +     [[ ! -d $ccachedir ]] && die "Invalid path '$passedccdir' given
> for ccache directory"
>
> die takes a format string and arguments. please don't inject code into
> the format string.
>

doesn't seem to work here - and is not done that in the script at all


> > +fi
> > +
> >  # Detect chrootdir filesystem type
> >  chroottype=$(stat -f -c %T "$chrootdir")
> >
> > @@ -349,6 +359,7 @@ download_sources
> >  if arch-nspawn "$copydir" \
> >       --bind-ro="$PWD:/startdir_host" \
> >       --bind-ro="$SRCDEST:/srcdest_host" \
> > +     ${ccachedir:+"--bind=$ccachedir:/build/.ccache"} \
>
> Why do you modify the ccachedir path? What guarantee is there that this
> extra path exists on top of the user provided directory? I'm not a
> ccache user, but this seems rather unexpected at a glance.
>

The build directory is created by makearchrootpkg and the
.ccache subdir is automaticly created by systemd-nspawn.

The paths are different, because nobody can't use root's
.ccache dir and also this allows using different .ccache
directories for different architectures.


>
> >       /chrootbuild
> >  then
> >       move_products
> > --
> > 1.8.3.4
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.archlinux.org/pipermail/arch-projects/attachments/20130730/cc76ee3b/attachment.html>