[RFC] proposed user doc for nested trees

[Alexander Belchenko]

Aaron Bentley пишет:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Alexander Belchenko wrote:
>> Aaron, I've faced with the non-relative branch references in my scmproj
>> plugin. I'm also used at the beginning similar scheme to your layout,
>> where actual nested components were lightweight checkouts of subbranches.
>>
>> Once you will move entire nested construct in the filesystem all
>> lightweight checkouts will start to point to wrong location, because
>> lightweight checkouts are using absolute paths to their branches.
> 
> I'm using checkouts because working trees should be able to move around.
>  merge, revert, mv, and others may change the position of a subtree.
> But if the subtree's branch reference was relative, moving around could
> break the reference.  So I don't think it would be a good idea to make
> those branch references relative.

On initial checkout how the bzr will build nested components? The 
relative paths of nested components will be stored somewhere (in the 
inventory?), is it right?

So when user will move nested components withing entire construct this 
info will be updated as well?

So why the code that will update relative position of nested component 
(in the inventory?) cannot update relative branch reference as well?

>> There is possible to "shit happens": user has moved old nested construct
>> to the other location and recreate fresh nested construct in the old
>> location. And then lightweight checkouts in the moved locations suddenly
>> will start point to the new nested construct.
>>
>> Do you think this is the right behavior?
> 
> It's the best compromise I can think of.

I'm sad to hear this.

> 
> Aaron
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
> 
> iEYEARECAAYFAkoJejMACgkQ0F+nu1YWqI1wiQCffvuGeCU1DG47Uyiy8UYOLf/S
> GuIAn2dWQ8RFzpN1RZ9MDkhvMi92JJdt
> =wPFC
> -----END PGP SIGNATURE-----
> 
>