Re: Mangled and unclear attributions in the Emacs history

[Eric S. Raymond]

Eric Hanchrow <address@hidden>:
> What Paul said.  Eric's own rule of thumb was that the result of the
> conversion should look as if git had been in use the entire time; that
> would imply that the "name at the time of contribution" would have
> been recorded.

That is a good argument.  But there is another kind of ID change that
I don't think has the same kind of weight as the intentional change
from "Michael" to "Thomas" Bushnell.  Consider the differences between

1.  Attribution to "Jan D. <address@hidden>" vs. 
"Jan Djärv <address@hidden>"

2. Attribution to "Richard Stallman" vs. "Richard M. Stallman".  

I don't think fixing these is problematic, because doing so does not
change the name the person was using at time of contribution; rather,
it restores parts of that name that were casually elided.

The reason I care about coalescing such cases is because I think it helps
everybody if (for example) "Jan Djärv" is a reliable single identity for
purposes of reputation measures, whether informal or formalized via
something like Ohloh Kudos rank.

"Thomas Bushnell" actually wants to have a different reputation, a
different social identity, from "Michael Bushnell".  That's the point
of taking a religious name.  "Jan D.", on the other hand, is not 
intended to be a different reputational identity than "Jan Djärv" (or
at least I assume it isn't Jan can correct that if he wishes).
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>