[arch-projects] [PATCH] archrelease: call 'svn copy' once for all files

[Lukas Fleischer]

On Fri, Sep 09, 2011 at 02:35:25PM -0500, Dan McGee wrote:
> On Fri, Sep 9, 2011 at 2:29 PM, Lukas Fleischer
> <archlinux at cryptocrack.de> wrote:
> > On Fri, Sep 09, 2011 at 03:24:31PM -0400, Dave Reisner wrote:
> >> On Fri, Sep 09, 2011 at 09:15:03PM +0200, Lukas Fleischer wrote:
> >> > On Fri, Sep 09, 2011 at 02:58:03PM -0400, Dave Reisner wrote:
> >> > > On Fri, Sep 09, 2011 at 08:44:12PM +0200, Lukas Fleischer wrote:
> >> > > > On Thu, Sep 08, 2011 at 04:38:28PM -0500, Dan McGee wrote:
> >> > > > > We don't need to invoke svn copy on each file; it accepts multiple
> >> > > > > arguments. This cut Allan's time releasing one patch-friendly package
> >> > > > > from 5 minutes to 2 minutes.
> >> > > > >
> >> > > > > Signed-off-by: Dan McGee <dan at archlinux.org>
> >> > > > > ---
> >> > > > >  archrelease |    5 ++---
> >> > > > >  1 files changed, 2 insertions(+), 3 deletions(-)
> >> > > > >
> >> > > > > diff --git a/archrelease b/archrelease
> >> > > > > index 2f6a563..7bb4c8a 100755
> >> > > > > --- a/archrelease
> >> > > > > +++ b/archrelease
> >> > > > > @@ -52,9 +52,8 @@ for tag in "$@"; do
> >> > > > >               svn add --parents -q "repos/$tag"
> >> > > > >       fi
> >> > > > >
> >> > > > > -     for file in "${known_files[@]}"; do
> >> > > > > -             svn copy -q -r HEAD "$trunk/$file" "repos/$tag/"
> >> > > > > -     done
> >> > > > > +     # copy all files at once from trunk to the subdirectory in repos/
> >> > > > > +     svn copy -q -r HEAD ${known_files[@]/#/$trunk/} "repos/$tag/"
> >> > > >
> >> > > > +1 from me. I wondered why we were looping over the single files in the
> >> > > > first place. However, if we copy from current HEAD anyway, I don't see
> >> > > > any reason to not just use `svn copy -q -r HEAD "$trunk/" "repos/$tag/"`
> >> > > > here.
> >> > > >
> >> > >
> >> > > +1 from me provided we quote the array expansion.
> >> > >
> >> > >   $ known_files=(foo "bar baz") trunk=/some/path
> >> > >
> >> > >   $ printf '%s\n' ${known_files[@]/#/$trunk/}
> >> > >   /some/path/foo
> >> > >   /some/path/bar
> >> > >   baz
> >> > >
> >> > >   $ printf '%s\n' "${known_files[@]/#/$trunk/}"
> >> > >   /some/path/foo
> >> > >   /some/path/bar baz
> >> >
> >> > Well, how is this superior to just copying the whole trunk directory?
> >>
> >> No idea. Will svn copy pick up files that aren't tracked? (packages,
> >> src/, pkg/, signatures...)
> >
> > Nope, shouldn't. Unless I missed something. We copy from current HEAD
> > and already committed to trunk before, so we shouldn't miss any files
> > when using plain "copy" here.
> 
> Because you'd be reverting the whole damn reason I made these changes
> in the first place- to prevent the silly extra commit. SVN will *not*
> let you svn rm a directory and then copy a new one in its place in the
> same commit.

Nope, what I wanted to suggest is `svn rm` the files and `svn copy` the
directory thereafter. Obviously, SVN doesn't have any option to copy the
content of a directory only, though... :/ So, +1 to your patch with
Dave's quoting fix applied :)

> 
> What you suggest is exactly what we used to do:
> http://projects.archlinux.org/devtools.git/commit/?id=8384ad849dfe308ed3e63e252785b9a6f80474f5
> 
> -Dan