[PATCH] Obfuscate Me Gently

[Brendan Cully]

(did you mean for this to go to me personally instead of the list?)

On Wednesday, 07 June 2006 at 09:53, Vadim Gelfer wrote:
> On 6/5/06, Brendan Cully <brendan at kublai.com> wrote:
> >on IRC Alexander Schremmer noted that UTF-8 characters in the Author
> >field of hgweb's changelog get mangled. The attached patch softens
> >obfuscate to only mangle characters < 128. Other alternatives might
> >include:
> 
> this does not look right to me. correct thing would be to obfuscate
> utf8 properly.

I suppose. But at the moment I don't think there's actually any hard
requirement that the data be in UTF-8 (although it is assumed). So I'd
rather not badly mangle the data if it happens to be 8859-1 or
something. In practise, mangling only 7-bit characters is probably
effective (especially since it automatically mangles the entirety of
email addresses).

> >* splitting author into person and email, and only obfuscating the
> >  email portion, which has to be 7-bit anyway IIRC.
> 
> there is no requirement from hg that email portion exist, so treating
> it special is wrong.

I meant if it were there. It would fall back to no mangling if it
weren't. But mangling 7-bit only should have an equivalent effect
without making any assumptions about the author field.