Re: Emacs-devel Digest, Vol 47, Issue 14

[David Kastrup]

"Eric S. Raymond" <address@hidden> writes:

> David Kastrup <address@hidden>:
>> "Eric S. Raymond" <address@hidden> writes:
>> 
>> > I second this recommendation.  I am not yet sure Mercurial is the
>> > overall best-of-breed, but it is the most accessible of the big three
>> > (git, bzr, hg) and seems to have the best UI design.
>> 
>> Before making a decision, one should try the candidates on the Emacs
>> repository data which will constitute our real workload.
>
> Agreed.  In fact, you may have solved a problem for me by pointing
> this out -- I've been casting around for a large real-world codebase to
> do comparative benchmarks against, and Emacs might do nicely.
>  
>> The Emacs history is very unusual in size and structure. 
>
> It's larger than most, certainly.  What do you mean by unusual 
> in "structure"?

Very long, linear branches, very small commit granularity, no rename or
merge information except in the logs.  For the final import, one will
want to extract this information in some manner or other (most but not
all of the currently missing information should have originated from
Miles' arch repository and can possibly be reclaimed from there, too).

-- 
David Kastrup