Joseph Myers <jos...@codesourcery.com>:
> On Wed, 26 Aug 2015, Eric S. Raymond wrote:
>
> > > Although a few accounts may have been deleted on sourceware, and
> > > conceivably account names could have been reused after such deletion,
> > > it's
> > > the parts of the history that come from the gcc2 repository (trunk until
> > > the start of EGCS, premerge-fsf-branch after that) where there's a
> > > greater
> > > risk of conflicts (as in, the two systems had completely separate passwd
> > > files).
> >
> > Noted.
>
> At least one such case confirmed. You'll need to use different author
> maps for different sets of commits.
>
> In the gcc2 repository, dje was Doug Evans (this includes commits on
> premerge-fsf-branch, not just the early part of trunk). In the repository
> that started as the EGCS repository, dje is David Edelsohn.
OK, this is going to get pretty ugly then. I do not know nearly enough
about the history of the repo to find these pairs myself.
I'm going to need two things:
1. A precise description of what SVN branches and date spans need to be
considered as separate namespaces for this purpose. When I say
"precise" I mean that I will need to be able to express them as
reposurgeon selection sets. It is OK for the description to use
branch names and either Subversion revnos or dates.
2. A list of exceptions. Each exception should be an ambiguous username
(such as dle) followed by a descroption of the human name it corresponds
to in different namespaces.
I gather, for example, that there will be a gcc2 set.
--
<a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>
