Re: [Tcl-gsoc] Tk Photo / BitTorrent client

[Andreas K. <and...@ac...>]

Karol "Zal" Zalewski wrote:
> Hi,
> 
>   As I wrote before, I am still interested in Tk Photo Image
> Manipulation (maybe merged with Tk Factor Photo Image Handling as Clif
> Flynt mentioned). Also, I am thinking about writing BitTorrent client in
> Tcl (alternative for Tk project). I have some questions which are linked
> with that projects.
> 
>   First of all, is there any mentor who wants to who wants to
> participate mentioned projects? It would be a problem for me, if no one
> wants to take care of them ;]

I am a possible mentor for the BitTorrent project. I have to admit that I do 
not know anything about the internals of the torrent protocol. I do know about 
the encoding it uses, and wrote a decoder/encoder for that [1].

[1] http://docs.activestate.com/activetcl/8.5/tcllib/bee/bee.html


>   Second question. Last year Tcl had 9 slots for GSoC, but there was
> much more ideas and students that wanted to participate. It is obvious
> that there is some priority list of tasks. I think that Tk Photo Image
> Manipulation (merged with factor and icon themes if needed) is quiet
> important. But what about BitTorrent client? Is there any chance to
> accept that project? Do you need it?

The torrent client is very likely a WIBNI (Wouldn't It Be Nice If we had this), 
and as such the other may prioritize it down. That however depends much on the 
number of applications we get. On the flip side, a project with a higher 
priority may be crowded by applications, of which only one will be chosen.


>   Last, but not least. When I say "interested" I mean, that I have
> basics in Tcl/Expect, I know the idea of BT (but I need to get and read
> protocol specification [1]) and I have 3-4 days to make draft of
> application. Can you tell me more about that concept? 

When I came up with the idea I envisioned that the result would be a library 
handling the protocol, with some API to control it, and also the ability to get 
feedback during the operation (via callbacks [2]). A GUI was not part of the 
vision, except for something minimal to allow testing. Getting such would be a 
bonus. However, as said, to me getting the basic functionality for the 
interaction with other torrect clients and seeds, and the ability to act as 
seed are the most important to me.

[2]	The callbacks could then be used by an unspecified UI to drive parts of
	it (logs, status lights, progressbars, that kind of thing).

 > I think, it is
> possible to make (very) simple BitTorrent client without GUI in less
> than three months using Transmission (especially libtransmission [2])
> source code as an example. And it will be a good way to make something
> more after GSoC.

Agreed.

Andreas