On Thursday 04 Apr 2013 16:54:12 [email protected] wrote: > Hello Cheatan and Toad, > > Chetan Hosmani <[email protected]> writes: > > I had discussed with toad and we were definitely interested in letting > > another student take up and make further progress in transport > > plugins. I hope that is still the case :) > > Matthew Toseland <[email protected]> writes: > > All sounds promising. Certainly it would be good to have another student > > working on transport plugins. > > Thanks Cheatan and Toad for the enthusiastic reply. Sorry for replying late. > the digests don't come in everyday and I spend most of yesterday either > packing or on the plane. This is the first moment that I got access to > Internet. > > Chetan Hosmani <[email protected]> writes: > > I am definitely interested in mentoring if freenet is okay with it. > Thank you for offering help and time for this. Now, that my pre-proposal > did not met with total boredom/disgust, I can go and have a peek at the > transport branch, and maybe come back with bunch of questions. > > Matthew Toseland <[email protected]> writes: > > Yes, we don't know if we're even in GSoC. If we are, he will need to > > fix a bug in Freenet or implement a small feature to show that he can > > code. > Why don't we go with what toad said and you give me some simple task but > related to the transports. Beside to see if I can code or not, I can get > my toes wet, see what's going on in your code and write a more relevant > application that makes sense. In that way if either Freenet or my > application get rejected, least it'll be a partial achievement toward my > goal.
Well, implementing stream transports is probably too much. Implementing a simple stego'ed UDP transport might be an idea? (E.g. with a custom header) I suspect even this will require a lot of fixes to the transports branch though, since it only has the basic UDP transport working at the moment ... Might be easier to try something simpler. Have a look at the bug tracker? > > I do C++ and python most of the time and Java isn't at the top of my > list of expertise yet, but between C++ and Python, I have been able to > find my way in the Java land whenever the team has required me to do > so. It shouldn't take a long time for me to gain my full speed and my > code tastes more like Java. If you are happy with C++ then Java should be no problem. > > Also, please bear with me at the beginning while naturally there will > be lots of questions and clarification. I can do both email or irc which > ever is easier for you. Please feel free to ask questions. Asking questions is good, the more people that understand the code the better - often they find things that are wrong. :) You can ask stuff here or on #freenet on irc.freenode.net (but bear in mind timezones). > > Chetan Hosmani <[email protected]> writes: > > I will try to come on IRC this week (mostly the weekend) and discuss > > this. I think there would have to a prerequisite task that needs to be > > completed for this project? Also the accepted organisations have not > > yet been announced? > > > I haven't been on irc regularly as the sole tor hidden service able to > log into freenet isn't available all the time and sometime too busy to > offer any service. I think my ISP at home has some problem with it, but > I'm going to find a more reliable solution soon. We use irc.freenode.net. We (devs) don't use a tor hidden service. A few people use FLIP, but it's slow. For talking to the wider anonymous community FMS and Sone are the best options. > > Matthew Toseland <[email protected]> writes: > > Re small packets, that's useful for UDP too - e.g. if we wanted to > > look like Skype, we'd need to figure out a way to divide stuff into > > very small packets and then do HMACs etc at a higher level. > > Anyway looking forward to this. > I haven't code for UDP communication before, but I don't think it should > be hard to adopt some fork of Stegotorus for UDP. Stegotorus chops the > packets to desirable size, encrypt, mac and give them to the steg module > for stegano. Then it reassembles them again on the other end. It also > take care of retransmission in case of lost packets and > acknowledgment. Right. We need reliable HMACs at a higher level - our current code has a 10 byte HMAC, but even that is going to be too long for e.g. a 27 byte packet (e.g. Skype-like). We already provide retransmission etc; most of what we need is already there, you can create big packets and chop them up in a plugin, or you can modify the core to encode stuff differently, both are fine with me although IMHO it would be good to have core support for really small packets. > > > Re TCP, Chetan's architecture is intended to include stream > > transports, but that part hasn't been finished yet. Certainly we need > > stream transports for all sorts of stego plugins. > Do you (Chetan+Toad) think finishing stream transports up would be a > appropriate load for a GSoC project? Chetan code you tell me where > (which file) I should look for the functions related to stream > transporns in your branch. You'd need to work quite closely with Chetan; there's plenty of room to expand the proposal, once you get streams working you can e.g. write more stream transports, and no doubt there are complicated issues in the core code related to e.g. bandwidth limiting. However, *if* Chetan has the time to mentor (at least several hours per week IMHO, especially in core code which can be quite complex to understand), this seems like a good idea. Separately, obviously I'd like to get Chetan's work merged soon. :) > > Looking forward to talk to everybody on this matter. > > Cheers, > Vmon
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Devl mailing list [email protected] https://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
