[tor-talk] torsocks is broken and unmaintained

Jacob Appelbaum jacob at appelbaum.net
Sun Nov 4 21:31:57 UTC 2012


Nick Mathewson:
> On Sat, Nov 3, 2012 at 11:23 PM, Matthew Finkel <matthew.finkel at gmail.com>wrote:
> 
>> On 11/03/2012 08:38 PM, Nick Mathewson wrote:
>>
>  [...]
> 
>>> Okay, sounds like we've got some enthusiasm.  Let's get started.  I
>>> volunteer to review commits and if people ask me to, and suggest that
>>> asking me to review stuff for a while might be a smart idea.  I just gave
>>> myself commit access to the git at git-rw.torproject.org repo too, in case
>>> that helps.  I am not planning to be a primary author here.
>>
>> Thanks for adding one more thing to your plate! I know Jake can handle
>> this but the more eyes we have looking at these initial changes the
>> better it'll be.
>>
>>>
>>> Given the amount of people asking us to apply and/or warning us that we
>>> mustn't apply particular patches, I'm going to suggest the following
>>> principles for a while:
>>>   * LET'S START MINIMAL.  Let's stick to doing only the very major
>> bugfixes
>>> and obvious fixes for at least the next release or two, so that something
>>> usable comes out.
>>
>> Agreed. To be honest, I haven't really looked at the code too much, so
>> I'll start diving into that in a bit. (If there isn't one already...I
>> haven't checked) Can we get a trac component added so we can track
>> progress and such?
> 
> 
> Done.  At some point we should migrate issues from google code, but IMO
> that's best done once we have something nontrivial to show for our efforts.
> 

I'm on the google code project, I'd like to close out each of the issues
and open new issues on Tor's trac. That will allow us to ensure each
reporter will receive a desired response.

> 
>>>   * NO ARCHITECTURAL ASTRONAUTICS. I'm always tempted when I come to a
>>> codebase for the first time to refactor the heck out of it.  Let's avoid
>>> doing that till we have a little experience with this codebase.  There
>>> isn't all that much here: let's
>>
>> Yes...let's! :)
>>
>> Was there supposed to be more to that sentence?
> 
> 
> Yeah; sometimes I start a sentence, then I think of something to write
> elsewhere and start another sentence, but then by the time I'm done with
> that one I don't remember the first sentence any more, so it
> 
> That one should end with "There isn't all that much code here; let's make
> sure we understand it pretty thoroughly before we complexify it in the name
> of some half-glimpsed vision."
> 

Agreed.

> 
> 
>>>   * LOVE MEANS GET TESTED. If at all possible, we should make this
>> codebase
>>> easier to test (right now it wants you to install before testing), and
>>> improve the coverage of the tests so that (if as people suspect) we're
>>> likely to break things on one platform when we fix them on another, we
>> can
>>> at least find out fast whether a patch works everywhere.
>>>
>>
>> Certainly sounds like a good idea. I'm going to have to familiarize
>> myself with some of the other *nix platforms it does/should support.
>> Just looking through the current issues on google code, for example, I
>> don't know the internals of OSX well enough *yet* to know if [1] is even
>> possible. But once we've compiled a list of all the current critical
>> patches, Debian and others (assuming such a list doesn't exist already),
>> then we start applying, testing, revising, etc. :)
>>
>> [1] https://code.google.com/p/torsocks/issues/detail?id=41
> 
> 
> Hm. Supposedly, it's _supposed_ to work on OSX.  It has a lot of code for
> OSX support.  I just tried it with curl on my osx laptop, and it seemed to
> work okay.
> 

It should work on OS X and if it doesn't, we likely have OS X related
bugs. :)

All the best,
Jake


More information about the tor-talk mailing list