Patchwork .gitignore: add /bitbake to the ignore list for external distributions

login
register
mail settings
Submitter Jason Wessel
Date May 9, 2012, 1:46 p.m.
Message ID <1336571192-15711-1-git-send-email-jason.wessel@windriver.com>
Download mbox | patch
Permalink /patch/27375/
State New
Headers show

Comments

Jason Wessel - May 9, 2012, 1:46 p.m.
External distributions based on the oe-core will typically include
bitbake in the top level directory.  The idea is to make it easy
for external distributions to easily assemble a distribution
with a pristine version of oe-core, add avoid the git untracked messages:

 # Untracked files:
 #   (use "git add <file>..." to include in what will be committed)
 #
 #       bitbake/

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
---
 .gitignore |    1 +
 1 file changed, 1 insertion(+)
Koen Kooi - May 9, 2012, 1:58 p.m.
Op 9 mei 2012, om 15:46 heeft Jason Wessel het volgende geschreven:

> External distributions based on the oe-core will typically include
> bitbake in the top level directory.

Really?

>  The idea is to make it easy
> for external distributions to easily assemble a distribution
> with a pristine version of oe-core, add avoid the git untracked messages:
> 
> # Untracked files:
> #   (use "git add <file>..." to include in what will be committed)
> #
> #       bitbake/
> 
> Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
> ---
> .gitignore |    1 +
> 1 file changed, 1 insertion(+)
> 
> diff --git a/.gitignore b/.gitignore
> index 04e36c5..366151c 100644
> --- a/.gitignore
> +++ b/.gitignore
> @@ -1,3 +1,4 @@
> +/bitbake

Provided we want bitbake in here, shouldn't it be 'bitbake' to match the rest of the file?
Marko Lindqvist - May 9, 2012, 2:07 p.m.
On 9 May 2012 16:58, Koen Kooi <koen@dominion.thruhere.net> wrote:
>
> Op 9 mei 2012, om 15:46 heeft Jason Wessel het volgende geschreven:
>
>> External distributions based on the oe-core will typically include
>> bitbake in the top level directory.
>
> Really?

 Don't know about distributions, but when I first tried

> . openembedded-core/oe-init-build-env
> bitbake core-image-minimal

 it complained that bitbake was not there, so it seems to be some sort
of default of openembedded-core.


 - ML
Jason Wessel - May 9, 2012, 2:15 p.m.
On 05/09/2012 08:58 AM, Koen Kooi wrote:
> 
> Op 9 mei 2012, om 15:46 heeft Jason Wessel het volgende geschreven:
> 
>> External distributions based on the oe-core will typically include
>> bitbake in the top level directory.
> 
> Really?
> 


Certainly I cannot speak for everyone, but I would like to "attempt" establish some kind of standard and this forum is probably the best place to discuss it.  We would like to be able to assemble pristine git controlled pieces that comprise what you find in the Yocto Project.  I believe that Richard Purdie and Jeff Polk discussed this in a different forum and I was trying to "clean up" the assembly so to speak. 


>>
>> diff --git a/.gitignore b/.gitignore
>> index 04e36c5..366151c 100644
>> --- a/.gitignore
>> +++ b/.gitignore
>> @@ -1,3 +1,4 @@
>> +/bitbake
> 
> Provided we want bitbake in here, shouldn't it be 'bitbake' to match the rest of the file?


According to the git documentation, "man gitignore", you want the slash for the following reason:

--- excerpt from man page ---
       ยท   If the pattern does not contain a slash /, git treats it as a shell
           glob pattern and checks for a match against the pathname relative
           to the location of the .gitignore file (relative to the toplevel of
           the work tree if not from a .gitignore file).
-----------------------------

In theory there should not be a reason we need to use a pattern match.

Cheers,
Jason.
Phil Blundell - May 9, 2012, 2:21 p.m.
On Wed, 2012-05-09 at 08:46 -0500, Jason Wessel wrote:
> External distributions based on the oe-core will typically include
> bitbake in the top level directory.  The idea is to make it easy
> for external distributions to easily assemble a distribution
> with a pristine version of oe-core, add avoid the git untracked messages:

Presumably any non-trivial external distribution is going to include
other metadata as well, in which case they're going to have to
edit .gitignore for themselves anyway.  So it doesn't really seem as
though having bitbake be mentioned there in oe-core buys much.

On the other hand, actually moving bitbake inside the oe-core repository
seems like an idea which might have some merit.  It's not obvious that
having it in its own tree really achieves anything other than making
release engineering slightly more difficult.

p.
Jason Wessel - May 9, 2012, 2:45 p.m.
On 05/09/2012 09:21 AM, Phil Blundell wrote:
> On Wed, 2012-05-09 at 08:46 -0500, Jason Wessel wrote:
>> External distributions based on the oe-core will typically include
>> bitbake in the top level directory.  The idea is to make it easy
>> for external distributions to easily assemble a distribution
>> with a pristine version of oe-core, add avoid the git untracked messages:
> 
> Presumably any non-trivial external distribution is going to include
> other metadata as well, in which case they're going to have to
> edit .gitignore for themselves anyway.  So it doesn't really seem as
> though having bitbake be mentioned there in oe-core buys much.
> 
> On the other hand, actually moving bitbake inside the oe-core repository
> seems like an idea which might have some merit.  It's not obvious that
> having it in its own tree really achieves anything other than making
> release engineering slightly more difficult.


It appeared to me that the bitbake was maintained by an entirely different group of people, but oe-core is completely tied to bitbake.  It might always be the case that it will still be a separate repository and perhaps this is a side effect of how the poky git vs the development of the Yocto Project are maintained, with the other work flow being to assemble a distribution.

Technically the poky git has the same issue, but you do not see it because of the way it is managed.  Ideally I would have liked to see that change as well, but I understand it is done that way for simplicity vs using git submodules or subgit trees with a tool like repo or others.

It would be interesting to hear a few more view points on this topic.

Jason.
Koen Kooi - May 9, 2012, 3 p.m.
Op 9 mei 2012, om 16:45 heeft Jason Wessel het volgende geschreven:

> On 05/09/2012 09:21 AM, Phil Blundell wrote:
>> On Wed, 2012-05-09 at 08:46 -0500, Jason Wessel wrote:
>>> External distributions based on the oe-core will typically include
>>> bitbake in the top level directory.  The idea is to make it easy
>>> for external distributions to easily assemble a distribution
>>> with a pristine version of oe-core, add avoid the git untracked messages:
>> 
>> Presumably any non-trivial external distribution is going to include
>> other metadata as well, in which case they're going to have to
>> edit .gitignore for themselves anyway.  So it doesn't really seem as
>> though having bitbake be mentioned there in oe-core buys much.
>> 
>> On the other hand, actually moving bitbake inside the oe-core repository
>> seems like an idea which might have some merit.  It's not obvious that
>> having it in its own tree really achieves anything other than making
>> release engineering slightly more difficult.
> 
> 
> It appeared to me that the bitbake was maintained by an entirely different group of people

Yes, Chris Larson and Richard Purdie. Those names sound familiar :)
Richard Purdie - May 9, 2012, 3:54 p.m.
On Wed, 2012-05-09 at 15:21 +0100, Phil Blundell wrote:
> On Wed, 2012-05-09 at 08:46 -0500, Jason Wessel wrote:
> > External distributions based on the oe-core will typically include
> > bitbake in the top level directory.  The idea is to make it easy
> > for external distributions to easily assemble a distribution
> > with a pristine version of oe-core, add avoid the git untracked messages:
> 
> Presumably any non-trivial external distribution is going to include
> other metadata as well, in which case they're going to have to
> edit .gitignore for themselves anyway.  So it doesn't really seem as
> though having bitbake be mentioned there in oe-core buys much.
> 
> On the other hand, actually moving bitbake inside the oe-core repository
> seems like an idea which might have some merit.  It's not obvious that
> having it in its own tree really achieves anything other than making
> release engineering slightly more difficult.

The one thing I will say in favour of the separate repositories is that
its helped make the API and separation between the two very clear. For
example, bitbake has no knowledge of many variables such as WORKDIR and
this is the way it should be.

My main concern with any merge would be creep of knowledge from one to
the other which I think lets us have clean and well structured code at
the moment with good separation of responsibilities.

Cheers,

Richard
Philip Balister - May 9, 2012, 4:10 p.m.
On 05/09/2012 10:07 AM, Marko Lindqvist wrote:
> On 9 May 2012 16:58, Koen Kooi <koen@dominion.thruhere.net> wrote:
>>
>> Op 9 mei 2012, om 15:46 heeft Jason Wessel het volgende geschreven:
>>
>>> External distributions based on the oe-core will typically include
>>> bitbake in the top level directory.
>>
>> Really?
> 
>  Don't know about distributions, but when I first tried
> 
>> . openembedded-core/oe-init-build-env
>> bitbake core-image-minimal
> 
>  it complained that bitbake was not there, so it seems to be some sort
> of default of openembedded-core.

Seriously, putting bitbake inside the oe-core checkout is silly.

I guess I need to send my patch to oe-core I use to stop the locatoin
being hardcoded by the script that comes with oe-core.

Philip
Jason Wessel - May 9, 2012, 4:14 p.m.
On 05/09/2012 11:10 AM, Philip Balister wrote:
>
> I guess I need to send my patch to oe-core I use to stop the locatoin
> being hardcoded by the script that comes with oe-core.

Certainly that is another very reasonable approach to the problem.

Assuming it doesn't affect performance (and I can't imagine that it
would), I'll gladly have the oe-core and bitbake at the same level in
the directory hierarchy, and then there is absolutely no need for the
patch to the .gitignore.

Thanks,
Jason.
Khem Raj - May 10, 2012, 5:26 a.m.
On Wed, May 9, 2012 at 9:10 AM, Philip Balister <philip@balister.org> wrote:
>
> I guess I need to send my patch to oe-core I use to stop the locatoin
> being hardcoded by the script that comes with oe-core.

that will be nice.

Patch

diff --git a/.gitignore b/.gitignore
index 04e36c5..366151c 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,3 +1,4 @@ 
+/bitbake
 *.pyc
 *.pyo
 build*/conf/local.conf