gnu-arch-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gnu-arch-users] reminder: winning smallish project


From: Tom Lord
Subject: Re: [Gnu-arch-users] reminder: winning smallish project
Date: Fri, 19 Sep 2003 17:53:42 -0700 (PDT)


    > From: Robert Collins <address@hidden>

    > Seriously though, for a UI to expose the diference how about:

    > : to remove a revisions-patch & contents from your tree.
    > replay --remove <revision>

    > : to reverse the code changes
    > replay --reverse <revision>

    > : to apply a changeset in reverse
    > dopatch --reverse


    > I'm not sure the above is right, 

I'd guess you're sure it's wrong :-)


What you called `replay --remove' is harmonious with `dopatch
--reverse'.

What you called `replay --reverse' is dissonant with `dopatch
--reverse'. 

And the pivot point here, `dopatch --reverse', is harmonious with
`patch(1)' who is older than us, has better insurance, and who I've
heard has a really tough and mean big brother.

(And, btw, ':' is a terrible character to use that way.)

I'm sure what you really meant is:

    * to reverse the code changes and remove the associated patch log
       entries:

        % tla replay --reverse <revision>


    * to reverse only the code changes, leaving the patch log as-is:
    
        % tla replay --opposite <revision>

    * to apply a changeset in reverse (which, if the changeset adds
      patch log entries, will remove them):

        % tla dopatch --reverse [...]

    * to apply a changeset in reverse EXCEPT -- without changing the 
      patch-log entries:

        % tla dopatch --opposite [...]

(Which amounts to almost exactly what you said except that this
version "smells more like Tom" :-)


    > just provoking discussion.
    > Key point: dopatch should be ignorant of the patch logs, so it doesn't
    > need a 'reverse but keep logs' concept. 

BZZT.  Thank you for playing.  No, I think it is a pretty cool feature
of arch that the patch-log changes associated with archive
transactions work out to be ordinary, non-special uses of
mkpatch/dopatch.   Therefore, mkpatch/dopatch _do_ need a special
option if you want to ask them to overlook the patch log.

Say, all this is fine, but, um...  who's going to take the 3hrs to
actually do it?

-t





reply via email to

[Prev in Thread] Current Thread [Next in Thread]