Testing branch summary

Matthew Burgess matthew at linuxfromscratch.org
Thu Jul 8 12:04:37 PDT 2004


My work last night seems to have stirred things up a little, for which I
apologise - I should have sent this last night to describe exactly *why*
I did what I did:

1) As Ian mentioned, the testing branch didn't exist at all, so we had
to use *something* as a base.  We only had 2 places to choose from i)
unstable ii) stable

2) As Ian also pointed out unstable was/is much closer to what I'd
already posted would become testing/6.0.  It therefore made sense *to
me* to use that as a base, and just remove the unwanted changes. The
other option would have meant much more work IMO.  I saw the
requirement for a testing branch as a fairly urgent one, so I took the
easiest/quickest route.

3) Jeroen raised some points regarding the development model.  He
outlined 2 possibile scenarios for continued development using the 3
tier setup.  i) testing branches from stable and merges from unstable
or ii) testing branches from unstable and is beat-up until it's stable. 
In an ideal world I would have used (i), but I didn't due to the points
made in (2) above.  I don't foresee there being too much drift between
the 3 branches once 6.0 is out the door, which will make it much easier
and more natural to use (i) in the future.

4) Jeroen also mentioned having separate development teams to work on
unstable & testing.  Whilst I don't think we have the resources to be
able to make such a split, I do believe that the editors can "switch
hats" and work on the different branches, making suitable changes to
each one as and when they see fit.

I think that just about covers it.  If I did indeed miss something, or
anyone wants further clarification then don't hesitate to pester me :)

Best regards,


More information about the lfs-dev mailing list