threads / patch / 17889

patch, 6 partsModify description file to say what this file is

Subject: [PATCH 1/6] Modify description file to say what this file is

## tl;dr

6 messages between Feb 19, 2009 and Mar 4, 2009. Diffs are folded; open one to read it.

replies: 5people: 3as markdown or json

John Tapsell· Feb 19, 2009, 07:36 UTC · lore

A lot of people see this message for the first time on the gitweb interface, where there is no clue as to what 'this file' means.

Signed-off-by: John Tapsell <johnflux@gmail.com>
---
 templates/hooks--update.sample |    5 +++--
 templates/this--description    |    2 +-
 2 files changed, 4 insertions(+), 3 deletions(-)

Show changes to 2 files +4 −3

templates/hooks--update.sample, templates/this--description

diff --git a/templates/hooks--update.sample b/templates/hooks--update.sample
index 93c6055..f753d28 100755
--- a/templates/hooks--update.sample
+++ b/templates/hooks--update.sample
@@ -43,10 +43,11 @@ allowdeletetag=$(git config --bool hooks.allowdeletetag)
 
 # check for no description
 projectdesc=$(sed -e '1q' "$GIT_DIR/description")
-if [ -z "$projectdesc" -o "$projectdesc" = "Unnamed repository; edit this file to name it for gitweb." ]; then
+case $projectdesc in "Unnamed repository;"*|'')
 	echo "*** Project description file hasn't been set" >&2
 	exit 1
-fi
+	;;
+esac
 
 # --- Check types
 # if $newrev is 0000...0000, it's a commit to delete a ref.
diff --git a/templates/this--description b/templates/this--description
index c6f25e8..dcde3c5 100644
--- a/templates/this--description
+++ b/templates/this--description
@@ -1 +1 @@
-Unnamed repository; edit this file to name it for gitweb.
+Unnamed repository; edit this file 'description' to name the repository

-- 
1.6.2.rc1.3.g7d31b.dirty

Junio C Hamano· Feb 19, 2009, 10:11 UTC · re: John Tapsell · lore

Re: [PATCH 1/6] Modify description file to say what this file is

[PATCH 1/6] Modify description file to say what this file is

Looks good.

[PATCH 2/6] Google has renamed the imap folder

Jeff already pointed out an obvious thinko; I could fix-up locally (just ask).

[PATCH 3/6] Improve error message for branching an existing branch

The extra sentence is useless noise to annoy users and make them shout "none of your business!" back to git.

I would probably get this error message "already exists." more from forgetting to say "-f" in this sequence:

    $ git branch -f pu next
    $ git checkout pu
    $ sh rebuild-pu-script

to rebuild pu on top of updated next, and "did you mean to checkout?" misses the mark by a kilometer.

[PATCH 4/6] Improve error message for git-filter-branch

Looks good, with Sverre's rewording would be better, which I could locally squash in. Needs signoff, which I could locally forge (just ask to fix-up and forge).

[PATCH 5/6] Change output "error: " to "Error: " etc

Jeff is right, and the patch is wrong.

[PATCH 6/6] Mention to the user that they can reorder commits

The placement of the new message does not feel right, as adding anything near "If you remove ... WILL BE LOST" will cloud out that message which is more important.

I think it should come near or perhaps even before Commands, if we were to add anything here.

But I am afraid that the proposed new message will hurt the clueless users more than it would help them.

The cheat-sheet at the top is not for learning what the command can do for the first time. It is there to remind people (who already have general idea on what can be done) how exactly the commands are spelled. If somebody does not even know that the purpose of rebase-i is to amend and resequence, he will more likely destroy his history by blindly using the command without knowing what is going on, than making a lucky guess.

For that reason, a more appropriate line to add, if we were to add anything, might be:

 #  s, squash = use commit, but meld into previous commit
 #
+# If you do not know what is going on, remove everything and exit the editor!
+#
 # If you remove a line here THAT COMMIT WILL BE LOST.
 # However, if you remove everything, the rebase will be aborted.

John Tapsell· Feb 19, 2009, 10:18 UTC · re: Junio C Hamano · lore

Re: [PATCH 1/6] Modify description file to say what this file is

2009/2/19 Junio C Hamano <gitster@pobox.com>:

Show 64 quoted lines

> [PATCH 1/6] Modify description file to say what this file is
>
> Looks good.
>
> [PATCH 2/6] Google has renamed the imap folder
>
> Jeff already pointed out an obvious thinko; I could fix-up locally (just
> ask).
>
> [PATCH 3/6] Improve error message for branching an existing branch
>
> The extra sentence is useless noise to annoy users and make them shout
> "none of your business!" back to git.
>
> I would probably get this error message "already exists." more from
> forgetting to say "-f" in this sequence:
>
>    $ git branch -f pu next
>    $ git checkout pu
>    $ sh rebuild-pu-script
>
> to rebuild pu on top of updated next, and "did you mean to checkout?"
> misses the mark by a kilometer.
>
> [PATCH 4/6] Improve error message for git-filter-branch
>
> Looks good, with Sverre's rewording would be better, which I could locally
> squash in.  Needs signoff, which I could locally forge (just ask to fix-up
> and forge).
>
> [PATCH 5/6] Change output "error: " to "Error: " etc
>
> Jeff is right, and the patch is wrong.
>
> [PATCH 6/6] Mention to the user that they can reorder commits
>
> The placement of the new message does not feel right, as adding anything
> near "If you remove ... WILL BE LOST" will cloud out that message which is
> more important.
>
> I think it should come near or perhaps even before Commands, if we were to
> add anything here.
>
> But I am afraid that the proposed new message will hurt the clueless users
> more than it would help them.
>
> The cheat-sheet at the top is not for learning what the command can do for
> the first time.  It is there to remind people (who already have general
> idea on what can be done) how exactly the commands are spelled.  If
> somebody does not even know that the purpose of rebase-i is to amend and
> resequence, he will more likely destroy his history by blindly using the
> command without knowing what is going on, than making a lucky guess.
>
> For that reason, a more appropriate line to add, if we were to add
> anything, might be:
>
>  #  s, squash = use commit, but meld into previous commit
>  #
> +# If you do not know what is going on, remove everything and exit the editor!
> +#
>  # If you remove a line here THAT COMMIT WILL BE LOST.
>  # However, if you remove everything, the rebase will be aborted.
>
>

Junio,

  Thanks - I like everything you said.  Could you go ahead and commit
the accepted ones, with all the fix ups mentioned?

Thanks!

JohnFlux

John Tapsell· Mar 4, 2009, 07:07 UTC · re: Junio C Hamano · lore

Re: [PATCH 1/6] Modify description file to say what this file is

Hi Junio,
  Did you commit my patches?  I don't see them yet in git.git ?

John Tapsell

2009/2/19 Junio C Hamano <gitster@pobox.com>:

Show 64 quoted lines

> [PATCH 1/6] Modify description file to say what this file is
>
> Looks good.
>
> [PATCH 2/6] Google has renamed the imap folder
>
> Jeff already pointed out an obvious thinko; I could fix-up locally (just
> ask).
>
> [PATCH 3/6] Improve error message for branching an existing branch
>
> The extra sentence is useless noise to annoy users and make them shout
> "none of your business!" back to git.
>
> I would probably get this error message "already exists." more from
> forgetting to say "-f" in this sequence:
>
>    $ git branch -f pu next
>    $ git checkout pu
>    $ sh rebuild-pu-script
>
> to rebuild pu on top of updated next, and "did you mean to checkout?"
> misses the mark by a kilometer.
>
> [PATCH 4/6] Improve error message for git-filter-branch
>
> Looks good, with Sverre's rewording would be better, which I could locally
> squash in.  Needs signoff, which I could locally forge (just ask to fix-up
> and forge).
>
> [PATCH 5/6] Change output "error: " to "Error: " etc
>
> Jeff is right, and the patch is wrong.
>
> [PATCH 6/6] Mention to the user that they can reorder commits
>
> The placement of the new message does not feel right, as adding anything
> near "If you remove ... WILL BE LOST" will cloud out that message which is
> more important.
>
> I think it should come near or perhaps even before Commands, if we were to
> add anything here.
>
> But I am afraid that the proposed new message will hurt the clueless users
> more than it would help them.
>
> The cheat-sheet at the top is not for learning what the command can do for
> the first time.  It is there to remind people (who already have general
> idea on what can be done) how exactly the commands are spelled.  If
> somebody does not even know that the purpose of rebase-i is to amend and
> resequence, he will more likely destroy his history by blindly using the
> command without knowing what is going on, than making a lucky guess.
>
> For that reason, a more appropriate line to add, if we were to add
> anything, might be:
>
>  #  s, squash = use commit, but meld into previous commit
>  #
> +# If you do not know what is going on, remove everything and exit the editor!
> +#
>  # If you remove a line here THAT COMMIT WILL BE LOST.
>  # However, if you remove everything, the rebase will be aborted.
>
>

Jeff King· Mar 4, 2009, 10:00 UTC · re: John Tapsell · lore

Re: [PATCH 1/6] Modify description file to say what this file is

On Wed, Mar 04, 2009 at 07:07:14AM +0000, John Tapsell wrote:

> Hi Junio,
>   Did you commit my patches?  I don't see them yet in git.git ?

Try looking in 'next', which is where new patches cook before going to master (especially now, as we are right in the middle of release freeze on master).

  git log --author=John.Tapsell origin/next

-Peff

Junio C Hamano· Mar 4, 2009, 10:01 UTC · re: John Tapsell · lore

Re: [PATCH 1/6] Modify description file to say what this file is

John Tapsell <johnflux@gmail.com> writes:

> Hi Junio,
>   Did you commit my patches?  I don't see them yet in git.git ?

Please look for them in the 'next' branch.

Patches that came after -rc1 are at best queued in 'next' for post 1.6.2. The only exceptions are critical and obvious fixes, and fixes to regressions introduced since 1.6.1.