On Fri, May 31, 2019 at 11:18 PM, Dominique Dumont <d...@debian.org>
wrote:
On Wed, 29 May 2019 10:36:27 +0500 Pirate Praveen
<prav...@onenetbeyond.org>
wrote:
justification: it should not remove any existing copyright noticed
added by maintainer.
Then what's the point of running "cme update dpkg-copyright" ?
To find out if we missed any copyright notices.
Let's see what's going on:
node-gulp$ licensecheck -r make-iterator --copyright -m
make-iterator/LICENSE MIT/X11 (BSD like) 2014-2018 Jon
Schlinkert.
make-iterator/README.md UNKNOWN 2012-2013 moutjs team and contributors
(http:moutjs.com)
make-iterator/index.js UNKNOWN 2014-2018 Jon Schlinkert.
make-iterator/package.json UNKNOWN *No copyright*
First problem: LICENSE and README.md do not contain the same copyright
owners. By reading the README.md file, I saw that make-iterator is
derived from moutjs. Hence debian/copyright entry is accurate.
But how can cme decide if the discrepancy is due to upstream change or
upstream inconsistencies ? It cannot.
Don't remove anything if cme is not sure. When more than one notice is
there, add both, I think that is safer. In this case both notices are
still present.
license-reconcile choose to throw an error in this case. cme trusts
upstream files.
README.md is also upstream file.
To avoid update debian/copyright with wrong entries, you should
override wrong copyright information in
debian/fill.copyright.blanks.yml as described in
Dpkg::Copyright::Scanner man page.
Note that fill.copyright.blanks can be edited with "cme edit dpkg"
That said, tests done with node-gulp has shown that the way cme
extracts information from LICENSE and README file is not ideal. I'm
going to improve its behaviour.
Thanks.
All the best.
