Dear Maintainer,

The regular expressions in sig2dot don't match the new output format
in gpg 2.1

The new output format is (using the same keys as in the script):

pub   dsa1024 2003-08-01 [SC]
      75D8908EBD35E31D94D102EF10557B83807CAC25
uid           [marginal] Michael Ablassmeier (abi) <abi@grinser.de>
sig 3        CE04080E9456ADE2 2004-02-07  Michael Schiansky <michael@schiansky.de>

Cheers,
Andrew

Hi,

I'm not sure whether it's worth to fix this: gpg's output without
--with-colons is designed to be human-readable, not machine-parseable.
Furthermore there is no guaranty that the format won't change across
versions (as you just pointed out :-P).  Passing the ‘--keyid-format’
option with a value other than ‘none’ will make sig2dot work again with
2.1.15.  But I think on the long run the proper fix would be to rewrite
sig2dot to make it read the colon-separated input produced by ‘gpg
--with-colons’.

Cheers,

Hi

I have not looked in detail, but there seem to be at least one rework
of sig2dot rewritten which seem to cover this.

It is https://github.com/bmhm/sig2dot2

Commits stop as well on beginning of 2019 though.

Regards,
Salvatore

Hi Salvatore!

Thanks for the pointer, I was not aware of this project.  On first
glance it seems it could replace the existing sig2dot version (at the
cost of an entry in the NEWS file), will certainly consider its
inclusion for the next release.

It's still an improvement over the old sig2dot, and not a dealbreaker if
it's already feature complete :-)  (Not sure about this, but I note that
open issues are harmless.)

cheers

For what it's worth, I'm working on writing a sig2dot or very similar tool in
C. Unlike sig2dot(2) it doesn't do text parsing of GnuPG's CLI output, which
is subject to change with the users' preferences and discouraged upstream, but
instead is to use the GPGME library to examine signatures.

I'm working on using the Graphviz library to generate the output as well. For
these reasons I hope it will be fast and handle extremely large keystores.