git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Marc Weber <marco-oweber@gmx.de>
To: git@vger.kernel.org
Subject: Re: [PATCH] topgit tg push feature
Date: Mon, 11 May 2009 05:28:13 +0200	[thread overview]
Message-ID: <20090511032813.GA15540@gmx.de> (raw)
In-Reply-To: <20090509190910.GA9655@gmx.de>

Feedback to Berts reply:

  > tg-push first. And where is the --dry-run option?
  fixed: added -h|--help option

  >This phrase needs rewording, what about: "its deps, both their tgish
  >and non-tgish ones."
  fixed, used Uwes text.

  It looks like this now:

          $tg push remote
          to push the current branch, its deps and, both their tgish
          and non-tgish ones.  You may add --no-deps and or --tgish-only
          to change this default behaviour. Use these only if you know what you
          are doing. They are only present for the sake of completeness.


  >now i see why you have the Usage: in the README. common practice is to
  >print the usage if an unknown option was given. see all the other
  >tg-*.sh scripts.
  fixed: see -h|--help above. tg export lists the Usage as well in the README. (-> other topic)



  >How effetcts the tg -r REMOTE option this command. Or more exactly why
  >doesn't have this option an effect here?
  laziness. I thought about that myself. Its fixed now. Usage looks like this:

    Usage: tg push [(--dry-run | --no-deps | --tgish-only)] [-r "remote remote2"]* branch*


  > the common error message is: "not a TopGit-controlled branch".
  :-) The point about this is: You can use
    $tg push -r "remote1 remote2" non-tgish-branch1 non-tgish-branch2.
  tg push will note that it can't find remotes but it'll push those branches.
  So you can "abuse" tg push to push normal branches to multiple remote
  locations.

  > why not pass the --dry-run option to git remote?
  git remote -> git push:
  I added --dry-run so that I can see the list of branches which would be pushed
  if I had not used --dry-run. It isn't that important that's why I missed adding
  it to Usage as well.

  
Feedback to Uwe Klein Koenigs mail:

  > I'd like to have a bit more prose.  Something like:
  fixed/misus


  > maybe extend the comment to something like:

  >         # if a remote is used ('topgit.remote' configuration variable or
  >         # tg -r $remote, a tg-base implicitly depends on
  >         # refs/remotes/$remote/top-bases/$_dep.  Don't push these.

  > You could even test if this is a dependency as described in this
  > comment.  Didn't test it, but something like

  >         test "x$_dep" = "xrefs/remotes/$base_remote/top-bases/$_name"

  > could do the trick.

  I've extended the recurse_deps function reading from the global var
  no_remotes now. It's still ugly but less verbose and more intuitive than
  adding long comments



  > > +       [ -z "$tgish_deps_only" ] || [ -n "$_dep_is_tgish" ] || return 0
  >I've always problems to understand these constructs.  Are these any
  >better than
  ... :-) *lol* You proposed using them :-)
  I do have a vary hard time reading writing them as well.
  I have to use boolean algebra to wrap my head around those constructs.


  > > +       echo "$_dep"
  > > +       local base="top-bases/$_dep"
  > > +       if ref_exists "$base"; then
  > > +               echo "top-bases/$_dep"
  > > +       else
  > > +               echo "warning, no base found $base" 1>&2
  > > +       fi
  >mmh, I wonder why you need to do the top-bases explicitly.  recurse_deps
  >doesn't that for you?
  Have a closer look at that function, please:
    "
      has_remote "top-bases/$_name" && [ -z "$no_remotes" ] &&
              echo "refs/remotes/$base_remote/top-bases/$_name" >>"$_depsfile"
      [..]
      ref_exists "refs/top-bases/$_dep"  ||
    "
  are all occurrences of top-bases. And both only do checks (or push remotes)



Feedback to Berts 2nd reply:

  >I also like the true/false style, its also useable with the current 'scheme':
  >    $tgish_deps_only && ...  
  great idea. true/false are sh builtins. adopted. I still prefer
  [ foo -a bar ]; or [ foo -o bar ]; for readability.

Sincerly
Marc Weber

commit cdab940eefc0a56e0daa8b270bace1aba00a3b57
Author: Marc Weber <marco-oweber@gmx.de>
Date:   Mon May 11 05:25:40 2009 +0200

    t/tg-push
    
    add tg-push  pushing the branch, its deps and their bases
    
    Signed-off-by: Marc Weber <marco-oweber@gmx.de>

diff --git a/.gitignore b/.gitignore
index eb56446..2f6d991 100644
--- a/.gitignore
+++ b/.gitignore
@@ -21,6 +21,8 @@
 /tg-import.txt
 /tg-remote
 /tg-remote.txt
+/tg-push
+/tg-push.txt
 /tg
 .*.swp
 
diff --git a/README b/README
index d2f095d..495c70b 100644
--- a/README
+++ b/README
@@ -480,6 +480,12 @@ tg update
 
 	TODO: tg update -a for updating all topic branches
 
+tg push
+~~~~~~~
+	pushes a TopGit-controlled topic branch to a remote
+	repository.  By default the remote gets all dependencies
+	(both tgish and non-tgish) and bases pushed to.
+
 TODO: tg rename
 
 
diff --git a/tg-push.sh b/tg-push.sh
new file mode 100644
index 0000000..0fa7854
--- /dev/null
+++ b/tg-push.sh
@@ -0,0 +1,82 @@
+#!/bin/sh
+# TopGit - A different patch queue manager
+# GPLv2
+
+remotes=
+
+## Parse options see README
+
+recurse_deps=true
+tgish_deps_only=false
+dry_run=false
+
+while [ -n "$1" ]; do
+	arg="$1"; shift
+	case "$arg" in
+	--no-deps)
+		recurse_deps=false;;
+	--dry-run)
+		dry_run=true;;
+	--tgish-only)
+		tgish_deps_only=true;;
+	-h|--help)
+		echo "Usage: tg push [(--dry-run | --no-deps | --tgish-only)] [-r "remote remote2"]* branch*"
+		exit 1;;
+	-r)
+		remotes="$remotes $1"
+		shift
+		;;
+	*)
+		branches="$branches $arg";;
+	esac
+done
+
+if [ -z "$remotes" ]; then
+	remotes="$(git config topgit.remote 2>/dev/null)"
+fi
+
+if [ -z "$remotes" ]; then
+	die "no remote location given. Either use -r remote argument or set topgit.remote"
+fi
+
+if [ -z "$branches" ]; then
+	branches="$(git symbolic-ref HEAD | sed 's#^refs/heads/##')"
+fi
+
+for name in $branches; do
+	ref_exists "$name" || die "detached HEAD? Can't push that"
+done
+
+push_branch(){
+	# if so desired omit non tgish deps
+	 $tgish_deps_only && [ -z "$_dep_is_tgish" ] && return 0
+
+	echo "$_dep"
+	local base="top-bases/$_dep"
+	if ref_exists "$base"; then
+		echo "top-bases/$_dep"
+	else
+		echo "warning, no base found $base" 1>&2
+	fi
+}
+
+for remote in $remotes; do
+	for name in $branches; do
+		list="$(
+			# deps
+			if $recurse_deps; then
+				no_remotes=1 recurse_deps push_branch "$name"
+			fi
+			# current branch
+			_dep="$name"
+			_dep_is_tgish=1
+			push_branch "$name"
+		)"
+		echo "pushing:"; echo $list
+		if $dry_run; then
+			echo git push $remote $list
+		else
+			git push $remote $list
+		fi
+	done
+done
diff --git a/tg.sh b/tg.sh
index 0804f73..94f38f4 100644
--- a/tg.sh
+++ b/tg.sh
@@ -136,6 +136,7 @@ branch_annihilated()
 # of the whole function.
 # If recurse_deps() hits missing dependencies, it will append
 # them to space-separated $missing_deps list and skip them.
+# set no_remotes to any value to omit remote dependencies (-> tg push)
 recurse_deps()
 {
 	_cmd="$1"; shift
@@ -145,9 +146,8 @@ recurse_deps()
 	_depsfile="$(mktemp -t tg-depsfile.XXXXXX)"
 	# Check also our base against remote base. Checking our head
 	# against remote head has to be done in the helper.
-	if has_remote "top-bases/$_name"; then
+	has_remote "top-bases/$_name" && [ -z "$no_remotes" ] &&
 		echo "refs/remotes/$base_remote/top-bases/$_name" >>"$_depsfile"
-	fi
 
 	# if the branch was annihilated, there exists no .topdeps file
 	if ! branch_annihilated "$_name"; then

  reply	other threads:[~2009-05-11  3:28 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-04-25 16:13 [PATCH] topgit tg push feature Marc Weber
2009-04-25 16:34 ` Marc Weber
2009-05-05  9:34   ` Marc Weber
2009-05-07  4:59     ` Marc Weber
2009-05-07  5:45       ` Bert Wesarg
2009-05-07  8:43         ` Uwe Kleine-König
2009-05-07  9:50           ` Bert Wesarg
2009-05-09 10:36             ` martin f krafft
2009-05-09 19:09               ` Marc Weber
2009-05-11  3:28                 ` Marc Weber [this message]
2009-05-11  7:25                   ` martin f krafft
2009-05-11  7:47                   ` martin f krafft
2009-05-11 19:55                   ` Uwe Kleine-König
2009-05-11 21:06                     ` Marc Weber
2009-05-11 21:39                       ` martin f krafft
2009-05-12  5:13                       ` Uwe Kleine-König
2009-05-12  6:22                         ` Bert Wesarg
2009-05-12  7:54                         ` Marc Weber
2009-05-12  8:55                           ` Bert Wesarg
2009-05-12  9:02                             ` Uwe Kleine-König
2009-05-13 10:04                               ` [PATCH] tg-remote: don't add push specs but warn about existing ones Uwe Kleine-König
2009-05-13 11:42                                 ` martin f krafft

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20090511032813.GA15540@gmx.de \
    --to=marco-oweber@gmx.de \
    --cc=git@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).