[PATCH] Tweak git-quiltimport to allow more flexible series format

[PATCH] Tweak git-quiltimport to allow more flexible series format

[Alexey Dobriyan]

Make quiltimport also understand comments following patch name.

Signed-off-by: Alexey Dobriyan <adobriyan@sw.ru>
---

 git-quiltimport.sh |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

--- a/git-quiltimport.sh
+++ b/git-quiltimport.sh
@@ -63,7 +63,7 @@ tmp_info="$tmp_dir/info"
 commit=$(git rev-parse HEAD)
 
 mkdir $tmp_dir || exit 2
-for patch_name in $(grep -v '^#' < "$QUILT_PATCHES/series" ); do
+for patch_name in $(sed -e 's/#.*//' < "$QUILT_PATCHES/series" ); do
 	if ! [ -f "$QUILT_PATCHES/$patch_name" ] ; then
 		echo "$patch_name doesn't exist. Skipping."
 		continue

Re: [PATCH] Tweak git-quiltimport to allow more flexible series format

[Junio C Hamano]

Alexey Dobriyan <adobriyan@sw.ru> writes:

> Make quiltimport also understand comments following patch name.
>
> Signed-off-by: Alexey Dobriyan <adobriyan@sw.ru>
> ---
>
>  git-quiltimport.sh |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> --- a/git-quiltimport.sh
> +++ b/git-quiltimport.sh
> @@ -63,7 +63,7 @@ tmp_info="$tmp_dir/info"
>  commit=$(git rev-parse HEAD)
>  
>  mkdir $tmp_dir || exit 2
> -for patch_name in $(grep -v '^#' < "$QUILT_PATCHES/series" ); do
> +for patch_name in $(sed -e 's/#.*//' < "$QUILT_PATCHES/series" ); do
>  	if ! [ -f "$QUILT_PATCHES/$patch_name" ] ; then
>  		echo "$patch_name doesn't exist. Skipping."
>  		continue

Is this consistent with the way quilt groks the series file?

IOW, does quilt forbid patchfile whose name contains a hash, and
anything after a hash on the line is taken as comment?

Can a line in a quilt series file name more than one patchfile?
If so, are they whitespace separated?

Re: [PATCH] Tweak git-quiltimport to allow more flexible series format

[Alexey Dobriyan]

On Thu, Nov 22, 2007 at 03:43:53PM -0800, Junio C Hamano wrote:
> Alexey Dobriyan <adobriyan@sw.ru> writes:
> 
> > Make quiltimport also understand comments following patch name.
> >
> > Signed-off-by: Alexey Dobriyan <adobriyan@sw.ru>
> > ---
> >
> >  git-quiltimport.sh |    2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > --- a/git-quiltimport.sh
> > +++ b/git-quiltimport.sh
> > @@ -63,7 +63,7 @@ tmp_info="$tmp_dir/info"
> >  commit=$(git rev-parse HEAD)
> >  
> >  mkdir $tmp_dir || exit 2
> > -for patch_name in $(grep -v '^#' < "$QUILT_PATCHES/series" ); do
> > +for patch_name in $(sed -e 's/#.*//' < "$QUILT_PATCHES/series" ); do
> >  	if ! [ -f "$QUILT_PATCHES/$patch_name" ] ; then
> >  		echo "$patch_name doesn't exist. Skipping."
> >  		continue
> 
> Is this consistent with the way quilt groks the series file?
> 
> IOW, does quilt forbid patchfile whose name contains a hash, and
> anything after a hash on the line is taken as comment?

It looks like hash inside a patch name is legal.
 
> Can a line in a quilt series file name more than one patchfile?

No. Which means we can probably do sed -e 's/^#//' -e 's/[ \t]#.*//'
... let me experiment bit more.

> If so, are they whitespace separated?