On Fri, 3 Oct 2008, Bluebie, Jenna wrote:
> I really don't like having the manual on git. I feel instead we should have
[...]
I wondered where you were going with that, given that git facilitates
collaboration, but
> Shoobie has been playing with shoes, finds the manual is blatantly wrong and
> the argument order for one of the drawing methods is different from what is in
> the manual. Shoes user has shoes installed, and some kind of code editor, but
> they have never used svn or git or anything like that because they're only
> just learning to code now.
raises very good points about usability. You point about Ruby's
docs is well taken, though I think that is probably more hindered by
the Japanese-English language barrier being so high (it would be easier
to get the core team to patch things if...).
There does seem to be a git-backed wiki
http://github.com/sr/git-wiki/tree/master
so this should be doable.
The points about the human side of collaboration that I just trimmed
for space are well made, too.
>
>
> Another model which I really like is on each section in the manual, have a
> small 'comment' link. When pressed, a window opens which asks for a name and a
> comment, akin to that on _why's tumblogs. This way the level of anonymity is
> up to the user and what they put in that box. Then _why would get an email and
Good point of which I'm in favour. I'm getting fed up of
registering for the right to edit, but I know why that happens, and
that does remain a concern. I don't want the amazingly productive
_Why flooded with comment spam. That killed the RubyGarden Wiki,
despite automated and manual attempts to clobber it all. But the
human contact side of this is good. I'm not sure how to make it
work effectively though. I'd suggest rate limiting it, for a start,
but then you'll get the odd genius who has 10 valid contributions to
make in as many hours. Blacklist and whitelists involve too much
admin to scale globally. The voting model of stackoverflow looks
useful, but I don't know how well it works, you still get rubbish on
slashdot. Is email authentication sufficiently anonymous?
HMAC against against the secret:
seed = email
(3 + rand(20).times do
seed = SHA512::Digest.new(seed).hex # or whatever the syntax is,
# i forget
end
for authentication? HMAC still has the key exchange problem, but
may be secure enough in that real transactions it doesn't get sent.
Hugh
