Is git mktag supposed to accept git cat-file input?

Hi,

I am doing some experimenting with git-mktag, and was looking into the format it expects on input.

Should this sequence of commands work?

kalki:[~]% mkdir /tmp/gittest; cd /tmp/gittest
kalki:[/tmp/gittest]% git init
Initialized empty Git repository in /tmp/gittest/.git/
kalki:[/tmp/gittest]% echo "test" > test
kalki:[/tmp/gittest]% git add test
kalki:[/tmp/gittest]% git commit -m "commit-test"
[master (root-commit) c0ae36f] commit-test
 1 file changed, 1 insertion(+)
 create mode 100644 test
kalki:[/tmp/gittest]% git tag -m "tag-test" tag-test
kalki:[/tmp/gittest]% git cat-file -p e619
object c0ae36fee730f7034b1f76c1490fe6f46f7ecad5
type commit
tag tag-test
tagger Anand Kumria <akumria@acm.org> Thu Oct 25 00:32:32 2012 +0100

tag-test kalki:[/tmp/gittest]% git cat-file -p e619 | sed -e's/tag-test/tag-test2/' > tag-test2 kalki:[/tmp/gittest]% cat tag-test2 object c0ae36fee730f7034b1f76c1490fe6f46f7ecad5 type commit tag tag-test2 tagger Anand Kumria <akumria@acm.org> Thu Oct 25 00:32:32 2012 +0100

tag-test2 kalki:[/tmp/gittest]% git mktag < tag-test2 error: char112: missing tag timestamp fatal: invalid tag signature file kalki:[/tmp/gittest]% git --version git version 1.7.9.5

The error message related to the timezone / timestamp being incorrect but I can't see what the problem is.

Any advice appreciated.

Please CC me as I may miss your reply.

Thanks, Anand

-- “Don’t be sad because it’s over. Smile because it happened.” – Dr. Seuss

On Wed, Oct 24, 2012 at 4:39 PM, Anand Kumria <akumria@acm.org> wrote:

Yes, with a slight tweak...

'-p' means pretty-print, i.e. produce a human-readable format. mktag supports the raw format. So you should invoke it like this:

      $ git cat-file tag e619

which should produce something like:

   object c0ae36fee730f7034b1f76c1490fe6f46f7ecad5
   type commit
   tag tag-test
   tagger Anand Kumria <akumria@acm.org> 1351121552 +0100

   tag-test

and is the format expected by mktag.

-Brandon

Ahh, unix time. Of course.

Thanks Brandon.

On 25 October 2012 01:18, Brandon Casey <drafnel@gmail.com> wrote:

Anand Kumria venit, vidit, dixit 25.10.2012 02:58:

That's the only difference *at the time being*, but this is not guaranteed. Really, as Brandon says: "cat-file -p" is pretty printing for human readability (which could be improved), and "cat-file <type>" is the raw format which is the content being hashed to the sha1.