Re: [dev] Surf assumes all SSL connections are good, which is bad

2010-02-10 Thread David Thiel
On 02/09/10 23:54, Alexander Surma wrote:
> Well, the connection is definitely encrypted. Regardless of a man in
> the middle or not ;)
> However - I see your point.
> My suggestion would be, that we allow yet another userscript to handle
> this. I for one do not care for verifying certificates. But for those
> who do, some kind of interface would be nice, woudln't it?

I think that if SSL is going to be supported, it should be supported
fully, within surf itself. One of the things I like about surf is that
it's actually usable without hacky user scripts, unlike, say, uzbl.
Otherwise, there's really no point; you're basically sending everything
in the clear, and SSL hasn't been implemented in any meaningful fashion.
And worse yet, surf *acts* like it's doing SSL when it isn't. Without
verification, surf is only usable for non-sensitive content -- you'd
have to be huffing enormous amounts of glue to even consider logging
into your bank with it.

Everyone hates the CAs, there's no argument there. I even think that
distributing a CA store with the browser itself is a potentially bad
move. But I personally consider the approach of combining "TOFU" and CA
cert verification is a pretty decent heuristic. Using TOFU-only would be
a first for a web browser, and I think that could be kind of neat. But
the least, just check an environment variable for a certificate store,
and if it's there, verify and turn the status bar green. Otherwise, it
stays red.



Re: [dev] Why use Mercurial?

2010-02-14 Thread David Thiel
On 02/14/10 09:08, Jacob Todd wrote:
> On Sun, Feb 14, 2010 at 04:04:54PM +0100, Preben Randhol wrote:
>> So what is your point? Everything has to be written in C? This is
>> mindbogglingly stupid. Not only is C; high maintenance, slow
>> development cycle, insecure but also extremely low level.  
> 
> What the fuck are you talking about? *How* is C high maintenance, how does C
> have a slow development cycle, and how on earth is it insecure, other than
> however insecure the programmer makes it? Would you rather use C++ with "nice"
> libraries like boost?

Another thing that is mindbogglingly stupid is arguing on the internet
about revision control systems or programming languages.