On Wed, Jan 8, 2014 at 4:21 AM, Alexander Dahl <post@lespocky.de> wrote:
Hei hei,

Am 2014-01-08 07:12, schrieb Jon Ringle:
> On Wed, Jan 8, 2014 at 12:56 AM, Alexander Dahl <post@lespocky.de> wrote:
>> This is exactly what we do, alas we do not use any ptxdist magic but two
>> self written shell scripts residing in another git repository "scripts"
>> included as submodule. Our rules/foo.make use one script for the get and
>> one for the extract stage. Those scripts make are called with two extra
>> variables from rules/foo.in as parameter. For each of our own packages
>> those describe clone path and which git entity (branch, tag, changeset)
>> to checkout after cloning. All these packages are cloned to
>> local_src/foo and build with cmake in local_src/foo-git. I assume all
>> this specific to our needs, but that's the way we do it. If you're
>> interested in the scripts, tell me.
>>
>>
> Yes, I would be interested to see your scripts and how you use them in your
> *.make and *.in :)

Okay I edited them a little and attached it. Note: the scripts itself
are put in a subfolder of the BSP here named "scripts".

I decided to use the get stage for cloning a git repo or checking if
it's correctly cloned and if fetching all branches and changesets
without updating the working copy at this stage. The other script for
the extract stage does the actual checkout then. It works. ;-)

Thanks!

How do you deal with creating a rule/kernel.make file that does this if I build Linux using my own tag? It seems that kernel may need some special handling

Jon