bug#42162: Recovering source tarballs

[Timothy Sample]

Hi Ludo,

Ludovic Courtès <ludo@gnu.org> writes:

> I hope you’re well.  I was wondering if you’ve had the chance to fiddle
> with Disarchive since the summer?

Sort of!  I managed to get the entire corpus of tarballs that I started
with to work (about 4000 archives).  After that, I started writing some
documentation.  The goal there was to be more careful with serialization
format.  Starting to think clearly about the format and how to ensure
long-term compatibility gave me a bit of vertigo, so I took a break.  :)

I was kind of hoping the initial excitement at SWH would push the
project along, but that seems to have died down (for now).  Going back
to making sure it works for Guix is probably the best way to develop it
until I hear more from SWH.

> I’m thinking there are small steps we could take to move forward:
>
>   1. Have a Disarchive package in Guix (and one for guile-quickcheck,
>      kudos on that one!).

This will be easy.  The hang-up I had earlier was that I vendored the
pristine-tar Gzip utility (“zgz”).  Since then I don’t think it’s such a
big deal.

(I wrote Guile-QuickCheck ages ago!  It was rotting away on my disk
because I couldn’t figure out a good way to use it with, say, Gash.  It
has exposed several Disarchive bugs already.)

>   2. Have a Cuirass job running on ci.guix.gnu.org to build and publish
>      the disarchive-db.

I’m interested in running Cuirass locally for other reasons, so I should
have a good test environment to figure this out.  To be honest, I’ve had
trouble figuring out Cuirass in the past, so I was dragging my feet a
bit.

>   3. Integrate Disarchive in (guix download) to reconstruct tarballs.

I had a very simple patch that did this!  It was less exciting when it
sounded like SWH was going to use Disarchive directly.  However, like I
wrote, making Disarchive work for Guix is probably the best way to make
it work for SWH if they want it in the future.

> WDYT?

This all will have to wait in the queue for a bit longer, but I should
be able to return to it soon.  I think the steps listed above are good,
along with some changes I want to make to Disarchive itself.


--Tim