[ccan] CCAN on GSoC

[Rusty Russell]

On Wednesday 26 March 2008 04:42:35 Ryan MCGarvey wrote:
> Some questions about the project:
>
> -What are you running on the back-end?

Hi Ryan,

   At the moment we simply have a static web page and a bzr repository.  We 
felt it important to start with some code snippets so we could get a handle 
on how the whole thing would work.

   One point of the GSoC is to build that infrastructure.  Parts could be 
stolen from CPAN (or others), and parts written from scratch.

> -Are you looking for an API written in C? (i.e. build a client/server model
> in C that can submit/database/etc various C programs)

We're certainly going to want a C client that can fetch things, mainly for 
updating code when there's a new version.  I already mentioned this to one 
candidate, that we could have a:
	ccan_update --try (moves old versions out the way)
	ccan_update --accept/--reject (once you've checked new versions are OK)

This is harder than it looks, because of namespacization (a program we have 
which adds ccan_ prefixes to code).

> -On the back-end, how are each of the projects databased?

Currently the backend is a bze repository.  Using a revision control system 
makes sense to me, since authors can conveniently use it to commit changes to 
their own code.

> -Would you propose the CGI be written in C?

Probably, but not religiously.  I imagine that much of the content will be 
static, generated by changes to the repository.  New code submission is one 
place which will need a CGI though.

Cheers,
Rusty.
PS.  With so many questions, I've set up a wiki, and I'll paste some of that 
there soon.