coping with unknown keywords on --status-fd

Daniel Kahn Gillmor dkg at
Fri Sep 16 22:39:20 CEST 2016

On Fri 2016-09-16 02:24:39 -0400, Neal H. Walfield wrote:
> Of course.  But, we could introduce something like: if a keyword is
> preceded by a '!' and the applications does not understand the
> keyword, the application should abort.

This is a lot like the "critical" flag in OpenPGP subpackets.  I
understand why it would be useful for upstream.

But if you want to reserve this feature, the syntax and semantics needs
to be decided as soon as possible, and clearly documented.  Otherwise,
you get consumers of the API implemented that don't know about the
criticality flag, which blithely ignore newly-introduced critical status
lines.  You cannot simply introduce the criticality feature with the
first status line that needs the criticality, right?

I'm in the process of doing a bunch of triage of consumers of this API
within debian, and complete+successful implementations are unfortunately

Having a concise overview of these sort of concerns for implementers to
review would be really helpful.  Should i advise these projects that
they should fail/abort on encountering !-prefixed unknown status lines
or not?

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 930 bytes
Desc: not available
URL: </pipermail/attachments/20160916/2b9eb6af/attachment-0001.sig>

More information about the Gnupg-devel mailing list