Add scripts/backport-support.sh

  This script is intended for maintaining the stable branches and
distribution backports.

2017-02-09  Florian Weimer  <fweimer@redhat.com>

	* scripts/backport-support.sh (latest_commit): New file.
  

I do not have a strong opinion here, but this workflow in the patch is
basically a simple git workflow. Do we really need backup this? Shouldn't
we just add this kind of documentation on wiki?

On 09/02/2017 13:35, Florian Weimer wrote:
> This script is intended for maintaining the stable branches and
> distribution backports.
> 
> 2017-02-09  Florian Weimer  <fweimer@redhat.com>
> 
> 	* scripts/backport-support.sh (latest_commit): New file.
> 
> diff --git a/scripts/backport-support.sh b/scripts/backport-support.sh
> new file mode 100644
> index 0000000..2ece7ce
> --- /dev/null
> +++ b/scripts/backport-support.sh
> @@ -0,0 +1,110 @@
> +#!/bin/bash
> +# Create a patch which backports the support/ subdirectory.
> +# Copyright (C) 2017 Free Software Foundation, Inc.
> +# This file is part of the GNU C Library.
> +
> +# The GNU C Library is free software; you can redistribute it and/or
> +# modify it under the terms of the GNU Lesser General Public
> +# License as published by the Free Software Foundation; either
> +# version 2.1 of the License, or (at your option) any later version.
> +
> +# The GNU C Library is distributed in the hope that it will be useful,
> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> +# Lesser General Public License for more details.
> +
> +# You should have received a copy of the GNU Lesser General Public
> +# License along with the GNU C Library; if not, see
> +# <http://www.gnu.org/licenses/>.
> +
> +# This script does not backport the Makefile tweaks outside the
> +# support/ directory (which need to be backported separately), or the
> +# changes to test-skeleton.c (which should not be backported).
> +
> +set -e
> +
> +export LC_ALL=C
> +export GIT_CONFIG=/dev/null
> +export GTT_CONFIG_NOSYSTEM=0
> +export GIT_PAGER=
> +
> +usage () {
> +    cat >&2 <<EOF
> +usage: $0 {patch|commit}
> +EOF
> +    exit 1
> +}
> +
> +if test $# -ne 1 ; then
> +    usage
> +fi
> +
> +command="$1"
> +
> +case "$command" in
> +    patch|commit)
> +    ;;
> +    *)
> +	usage
> +	;;
> +esac
> +
> +# The upstream branch to work on.
> +branch=origin/master
> +
> +# The commit which added the support/ directory.
> +initial_commit=c23de0aacbeaa7a091609b35764bed931475a16d
> +
> +# We backport the support directory and this script.  Directories need
> +# to end in a /.
> +patch_targets="support/ scripts/backport-support.sh"
> +
> +latest_commit="$(git log --max-count=1 --pretty=format:%H "$branch" -- \
> +  $patch_targets)"
> +
> +# Simplify the branch name somewhat for reporting.
> +branch_name="$(echo "$branch" | sed s,^origin/,,)"
> +
> +command_patch () {
> +    cat <<EOF
> +This patch creates the contents of the support/ directory up to this
> +upstream commit on the $branch_name branch:
> +
> +EOF
> +    git log --max-count=1 "$latest_commit"
> +    echo
> +    git diff "$initial_commit"^.."$latest_commit" $patch_targets
> +    echo "# Before applying the patch, run this command:" >&2
> +    echo "# rm -rf $patch_targets" >&2
> +}
> +
> +command_commit () {
> +    git status --porcelain | while read line ; do
> +	echo "error: working copy is not clean, cannot commit" >&2
> +	exit 1
> +    done
> +    for path in $patch_targets; do
> +	echo "# Processing $path" >&2
> +	case "$path" in
> +	    [a-zA-Z0-9]*/)
> +		# Directory.
> +		git rm --cached --ignore-unmatch -r "$path"
> +		rm -rf "$path"
> +		git read-tree --prefix="$path" "$latest_commit":"$path"
> +		git checkout "$path"
> +		;;
> +	    *)
> +		# File.
> +		git show "$latest_commit":"$path" > "$path"
> +		git add "$path"
> +	esac
> +    done
> +    git commit -m "Synchronize support/ infrastructure with $branch_name
> +
> +This commit updates the support/ subdirectory to
> +commit $latest_commit
> +on the $branch_name branch.
> +"
> +}
> +
> +command_$command
>
  

On 02/09/2017 11:35 PM, Adhemerval Zanella wrote:
> I do not have a strong opinion here, but this workflow in the patch is
> basically a simple git workflow. Do we really need backup this? Shouldn't
> we just add this kind of documentation on wiki?

I think having this in a script ensures that the backports are consistent.

We also may have to blacklist backporting certain wrappers which use 
functionality which is not available on older branches.

Thanks,
Florian
  

* Florian Weimer:

> On 02/09/2017 11:35 PM, Adhemerval Zanella wrote:
>> I do not have a strong opinion here, but this workflow in the patch is
>> basically a simple git workflow. Do we really need backup this? Shouldn't
>> we just add this kind of documentation on wiki?
>
> I think having this in a script ensures that the backports are consistent.
>
> We also may have to blacklist backporting certain wrappers which use 
> functionality which is not available on older branches.

Any further comments?  Thanks.
  

On 02/10/2017 05:03 AM, Florian Weimer wrote:
> On 02/09/2017 11:35 PM, Adhemerval Zanella wrote:
>> I do not have a strong opinion here, but this workflow in the patch
>> is basically a simple git workflow. Do we really need backup this?
>> Shouldn't we just add this kind of documentation on wiki?
> 
> I think having this in a script ensures that the backports are
> consistent.

I agree.

I wish we had more scripts in the tree that helped all of the developers,
providing example workflows and tooling to facilitate that.

I don't think that saying "it's just basic git" is any reason not to have
the scripts checked in.

Just like Joseph's script 'build-many-glibcs.py' is just basic toolchain
building workflow, it is still immensely valuable for building things.
 
> We also may have to blacklist backporting certain wrappers which use
> functionality which is not available on older branches.

Agreed.

I think it's a logical and positive step for maintainers to share scripts.
  

On 20/02/2017 10:13, Florian Weimer wrote:
> * Florian Weimer:
> 
>> On 02/09/2017 11:35 PM, Adhemerval Zanella wrote:
>>> I do not have a strong opinion here, but this workflow in the patch is
>>> basically a simple git workflow. Do we really need backup this? Shouldn't
>>> we just add this kind of documentation on wiki?
>>
>> I think having this in a script ensures that the backports are consistent.
>>
>> We also may have to blacklist backporting certain wrappers which use 
>> functionality which is not available on older branches.
> 
> Any further comments?  Thanks.
> 

It is not a blocker, but I still think this kind of script should be on
an external wiki.
  

On 09 Feb 2017 16:35, Florian Weimer wrote:
> +export LC_ALL=C

why ?  seems like it'll mishandle UTF8 ?

> +usage () {

we seem to be inconsistent, but i prefer no spaces before () ...

> +    cat >&2 <<EOF

style is weird.  what's with the 4 space indent ?
we use two spaces normally.

> +if test $# -ne 1 ; then

`test` instead of `[` ?

i think we normally omit the space before the ;

> +case "$command" in
> +    patch|commit)
> +    ;;
> +    *)
> +	usage
> +	;;
> +esac

now we mix tabs ?

> +# Simplify the branch name somewhat for reporting.
> +branch_name="$(echo "$branch" | sed s,^origin/,,)"

branch_name=${branch#origin/}

> +    git status --porcelain | while read line ; do

this creates a subshell.  you really want:
  while read -d $'0' -r line; do
    ...
  done < <(git status --porcelain)

should also have `local line`
-mike