Thanks for this description and being careful about us.
I suggest to do like websites : automatic publication with no manual action. It is the
easiest and most up to date.
But we need to understand you process, to know when content is moving and when it is
stable, each letter you change is a correction we have to do in all languages. We are many
people, but not many per language team.
----- Reply message -----
De : "Zach Oglesby" <zach(a)oglesby.co
Pour :
<docs(a)lists.fedoraproject.org
Objet : Translation needs clarity on
how to get updates published and the process.
Date : ven., août 12, 2016 19:18
On Thu, Aug 11, 2016, at 09:34 AM, Brian Exelbierd wrote:
This email is to drive some discussion around $subject. It follows
from
a blog soon to be posted on the Fedora Community blog
from that blog:
Translation needs clarity on how to get updates published and the
process.
This seemed like a communication problem between the two projects
that
needed to be resolved with better docs on the process and hand-off
procedures. Because the tooling proposal will hopefully include
continuous deployment, this may become a lot easier in the future.
Please reply here for discussion.
regards,
bex
One of the goals of the new system is to handle publishing automatically, this will
include translations that are ready to be published. Unfortunately, we do not have the
final plan worked out yet, and I don't have a clear way to tell the translation team
how it will definitively work, but I can explain my the idea I have worked out in my head
and hope that it will help as we move forward.
Step 1: Commits are made to a release branch of a document
Step 2: The CI/CD system will run and create PO files push them to zanata
Step 3: Translation team works on string in zanata as they do now
Step 4: When a translation is ready to be published it is added to the configuration file
Step 4: The config change is checked into the release branch and the CI/CD system is
kicked off again and the new translation is added to the site.
The only manual step in this whole process is to add the translations to the config once
they are ready, but unfortunately I don't see a way past that to make sure we are not
publishing translations that are not done or have not even been started. This step can be
done a few ways. We can give translation team members access to the docs repos or we can
use the pull request feature of Pagure, either way this is a much cleaner process than
what he have now as a person only needs to be envolved one time in the build/publishing
process to include a new translation.