-----BEGIN PGP SIGNATURE-----
-----END PGP SIGNATURE-----
On 09/20/18 12:37, Mattia Rizzolo wrote:
>> work to integrate 'update-output-helper' and to generally "parse"
>> update_output.txt and get users more information
> Well, IME that's when most of annoying issues come into being.
> Reading update_excuses.html is very easy; sure it may be handier to not
> have to load the whole page each time, but it's too bothersome. OTOH
> understanding update_output.txt is much harder, and often one needs to
> use other tools (dose, …) to understand the failure reasons.
I agree. In fact, we started a discussion on that with britney upstream,
to figure out a way of getting a machine-readable format of
update-output.txt. I suggested YAML, because that's easy to parse and
also fairly easy for a user to read.
That said, we do already catch some of the common issues: missing
builds, unresolvable dependencies, packages that require MIRs, etc. What
we can't display are required rebuilds of reverse-dependencies. That's
where integrating update-output-helper would help, too.
Much of the issue here is to keep anything that britney has to generate
to be as simple as possible and not inject too much logic that would
slow down its processing or use up more memory -- britney has enough to
do already. In that sense, I thought of just massaging the output in
text-mode it currently does, to just add enough whitespace to the
printfs to pass things as YAML (or you know, the right characters in any
Mathieu Trudel-Lapierre <email@example.com>
Freenode: cyphermox, Jabber: firstname.lastname@example.org
4096R/65B58DA1 818A D123 0992 275B 23C2 CF89 C67B B4D6 65B5 8DA1