[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Gnu-arch-users] history of "forked" files
From: |
Tom Lord |
Subject: |
Re: [Gnu-arch-users] history of "forked" files |
Date: |
Wed, 21 Jan 2004 17:30:21 -0800 (PST) |
> From: Brian May <address@hidden>
> Tom> What do you mean by "have a.c's history as part of its own?"
> Tom> For example, suppose that patch-N predates the creation a1.c
> Tom> and a2.c and that you 'tla replay --reverse patch-N': do you
> Tom> want changes to a.c to be removed from _both_ a1.c and a2.c?
> Tom> or do you only mean: "If I prepare a report that shows all
> Tom> the changes ever made to a2.c, it will also show changes to
> Tom> a.c?"
> subversion has a "copy" command that will copy the file with all of
> its history, ie. create a branch of the file.
Subversion also does not have a coherent notion yet of how to handle
history-sensative-merging and it seems to me that the semantics of
that "copy" command are one of the areas where they have work to do.
As far as I can tell, there is a choice between having a system that
handles merging well and having a system that has the current
semantics of the subversion copy command. Subversion and arch make
this choice differently.
As I remarked earlier: for reporting purposes, you can easily add
custom log headers to record that a copy of a file is related to some
older file.
> I think you could do this in arch(?) by somehow declaring a2.c was
> derived from a.c in its initial version.
With a custom header.
> I am not convinced myself that it would be worth the effort.
> Why not just create a new a2.c file and rename a.c to a1.c?
That's where I am on this issue as well.
-t