Re: --events-fd, deb822 format (Re: [patch] always send the full name on the --status-fd)
[ Adding deity mailing list to the CC, in case there's someone else
there that wants to comment on this. For context, please see
On Fri, 2013-11-08 at 18:19:31 +0200, Eugene V. Lyubimkin wrote:
> 2013-11-07 08:40, Guillem Jover:
> > Eugene, this is about adding a new option to replace --status-fd,
> > currently proposed as emitting deb822 format messages, I'd like your
> > input on this.
> The idea, format and proposed patch look good to me. The only things
> popped in my head are:
> a) to specify whether all the values are guaranteed to be one-line, or what
> line continuations should be suppored;
Ah yes, I noticed the line-wrapping issue after having sent the patch,
the errors should probably be normalized in the same way they are for
status-fd. Otherwise we'll need to prepend spaces and separate blank
lines with « .» or similar. Michael, or was your proposed error format
I forgot to mention that given that the messages are currently
localized, it might make sense to use Message-L10n or similar to make
that clear, and to possibly allow non-localized messages in the future.
OTOH I think apt is currently interested in the non-localized messages
only to be able to filter some of them, so maybe doing that would be
trying to fix the wrong thing?
> b) to specify the format of <package-name> (an example suggests it would
> be always fully qualified "package:architecture", but explicitly saying
> this would be nice);
Also forgot to mention this on my previous mail. Would you prefer
having it arch-qualified or the information properly split into
another Architecture field?
> c) 'Event: warning' could be nice to have in addition to 'Event: error'.
Oh yeah, I'll think how to plug that and not affect the rest of the
code base using the warning functions.
I'll also try to clarify a bit more the format in the documentation.
Eugene (and the others too), any more information you think might make
your life easier? Either new events or fields?