AW: obliterate functionality?

Christian Boos cboos at neuf.fr
Wed Mar 19 05:00:21 CDT 2008


Jens.Wulf at sew-eurodrive.de wrote:
>> But this will invalidate a bunch of SHA-1 hashes, right? (Not just
>> the obliterated ones).
>>     
>
>   
>> I was thinking about ways to not invalidate previously
>> published hashes -- if that's possible.
>>     
> Create a hash collision by adding a special dummy file to each changeset. That means using lots of computing power (the better the hash, the more you need. Hopefully it is impossible with reasonable computing power). People will hate you for implementing this, as changeset hashes become useless for referring to a certain version of a project once this 'feature' is in use.
>   

If I'm right, nobody mentioned the 
http://www.selenic.com/mercurial/wiki/index.cgi/TrimmingHistory page so far.
I think it contains a relevant discussion of the problem at hand (and a 
more realistic solution than the one suggested above ;-) ).

-- Christian


More information about the Mercurial mailing list