Re: git clone through http

[Samuel Bronson <naesten at gmail dot com>]

Simon Marchi <simon.marchi@ericsson.com> writes:

> Hello,
>
> Cloning the binutils-gdb repository using the http address
> (http://sourceware.org/git/binutils-gdb.git) doesn't seem to work.
>
> $ git clone http://sourceware.org/git/binutils-gdb.git
> Cloning into 'binutils-gdb'...
> error: Unable to get pack index
> http://sourceware.org/git/binutils-gdb.git/objects/pack/pack-40a278cd270f8215e6b993869d659b2aff311eb8.idx
[...]

Such messages indicate that, in order for http access to work, someone
either needs to:

  1. Rename/copy the file "hooks/post-update.sample" in the repository
     to "hooks/post-update" to allow use of the "dumb" git protocol.
     (In case the file is missing, or for the lazy but still curious, I
     have also appended a copy to this message.)

or

  2. Somehow set up a "smart" web-based git server.  This seems to
     involve git-http-backend(1); for Apache, there are directions at
     <http://sethrobertson.github.io/HowToPutGitOnTheWeb/GitOnTheWeb.html#aro>
     (and earlier parts of that page).

Choice 2 is clearly more complicated and more httpd-specific, but also
more efficient in most cases.

-- 
Hi! I'm a .signature virus! Copy me into your ~/.signature to help me spread!

#!/bin/sh
#
# An example hook script to prepare a packed repository for use over
# dumb transports.
#
# To enable this hook, rename this file to "post-update".

exec git update-server-info