The [[design documentation|contribute/design/translation_platform]]
documents how the production setup works.
This page is about ideas that are not implemented (yet?).
# Responsibilities of the _Translation platform maintainers_
Here we dump ideas that could be useful when defining [[this
## In scope
- Upgrade Weblate to new upstream releases.
- Weblate crashes.
- Removing stuff from the Weblate database because the webapp got mad.
## Out of scope
- Work that requires sysadmin privileges or belongs to "sysadmins
provide an OS with the required packages and system services", e.g.
MariaDB is crashing
UX for new translators
[[!tails_ticket 16974]] tracks progress on this front.
This list of problems was built thanks to usability testing at
MozFest 2019. The "users" referred to below are all people who tried
to use Weblate for the first time. All these users were rather
technical. They were using a broad variety of devices (about half
smartphones and half laptops) and browsers (a couple Tor Browser, some
Firefox, some the default browser of their mobile OS).
The initial set of hypotheses and proposals below are intrigeri's.
- The _Suggested translations (10)_ link in the top bar is confusing:
Several users thought it was about suggestions of translations for
specific strings, that should be reviewed. It took them a while to
realize that this was not the case. But then it was still not clear
what this link is about, nor how to actually list the translation
suggestions that need to be reviewed.
- Nobody understood what the numbers next to the links in the top bar
meant. After asking them what they thought it could be, some
ventured hypothesis: total number of strings, number of untranslated
strings, number of languages.
And regarding the _Suggested translations (10)_ link, several users
assumed that 10 was the number of translation suggestions that
should be reviewed, which is incorrect.
- Several users found it hard to find what needs work and what they
should focus on first. Some specific issues in this area:
- Nobody clicked on the "core pages" link.
Hypothesis: the fact that link is in the middle of 8 other links
may not help.
Proposal: remove the "anonymous internet", "first steps", "Install
& upgrade", and "persistence" links from the top bar. It's not
clear to me whether these component lists are maintained (e.g.
lots of the install and upgrade stuff has moved in the last year;
are the new components listed there?).
- One user wondered what to translate first once core pages
- On the page that gives an overview of a component for a given
language, such as [this
there's a _Translate_ button, even when the component is fully
translated already. One user found it confusing to click
_Translate_ and land on a string that is already translated.
- One user would have liked to be able to give feedback on suggested
translations, e.g. via a comment, even though they were not
a reviewer and thus could not approve the suggestion.
- All users but one found the _Register_ button without needing help.
The other user, after reading the prominent "Please get in touch to
start translating Tails website:
https://tails.boum.org/contribute/how/translate/" message on the
Weblate homepage, clicked on that link and then got totally lost.
Proposal: instead directly point to [[a more focused doc
"Please get in touch", and rephrase this message.
- Several users did not manage to register an account because Weblate
told them "too many registration attempts" on their first attempt.
Some were using Tor, some their mobile data, some the shared Wi-Fi
provided by the event.
Hypothesis: the root cause of this problem looks like a sysadmin or
Weblate bug rather than a UX design problem. Maybe Weblate has
a rate limiting mechanism to protect against too many account
registration attempts in a given time span, regardless of what
device/IP is used?
- During the account registration process, one user commented that the
arithmetic CAPTCHA they were facing (7×8) felt too hard.
They managed to solve it eventually.
Hypothesis: even if the user can fallback to using a calculator,
having to do so can be damaging for their self-confidence and desire
to continue. The message we're sending here is even worse than that,
given the skills required to solve this CAPTCHA are entirely foreign
to the skills one needs to translate Tails.
Proposal: whatever blocks spammers is good enough. Can we configure
the difficulty of this CAPTCHA? For example, maybe basic addition
and subtraction between numbers up to 5 might be an OK trade-off.
- One user found it hard to discover the hamburger menu on
- A UX designer who attended the session gave feedback and
recommendations about the [account registration
What follows was freely transcribed and interpreted by intrigeri
from very rough notes and memory.
It's all about the first sentence ("By registering you agree to use
your name and email in version control system commits and provide
your contribution under license defined by each translated
project"), which has several problems:
- It uses technical jargon ("version control system commits") and
vague terms and conditions ("under license defined by each
translated project", that is?) ⇒ it may not be clear to the user
what they are agreeing with exactly. That's not a good way to ask
if they consent.
- It's outside of the main form so it's easy to skip. If we really
need users to explicitly agree with these conditions, then this
should be a checkbox right on top of the _Register_ button.
- It does not encourage new translators to continue.
Recommendation: add an introduction text that briefly explains why
registering is useful, thanks the visitor, and encourages them
Documentation for translators
This is feedback and recommendations from a UX designer about our
[[documentation for translators|contribute/how/translate]], freely
transcribed and interpreted by intrigeri from rough notes.
Proposals are intrigeri's, recommendations are this UX designer's.
- The list of teams makes the whole page longer than it could be.
It's not clear if the list of teams is up-to-date: it does not match
the languages that are enabled in Weblate.
Proposal: ask on -l10n@ what these pages are still useful for in the
- It's not clear what translators should do with the information found
in the "Tier-1 languages" section.
- A new translator first needs to skip the "For native English
speakers" section, that presumably is of little interest to the vast
majority of folks who landed on this page.
Recommendation: move this section further down the page.
Implemented in [[!tails_gitweb_commit 3317eea94a]].
- To be more encouraging, we should start by thanking new translators.
Implemented in [[!tails_gitweb_commit b2a0d9c495]].
- The page lacks structure which makes it harder to find the
information a new translator needs.
Hopefully fixed in [[!tails_gitweb_commit eb17bf9da4]]
- The "Translate this website" section has a number of problems:
- It's long and it's hard to find _the_ link most readers will need,
i.e. "on the translation platform", because there are 3 links in
the paragraph where it lives.
Hopefully fixed in [[!tails_gitweb_commit 20090cdd20]] and [[!tails_gitweb_commit 77bbbb2c42]].
- It contains obsolete information: "you need to get into contact
with a language team if you want to participate".
Hopefully fixed in [[!tails_gitweb_commit 69702d9faa]].
A hybrid approach
The Tails infrastructure uses Puppet to make it easier to enforce and
replicate system configuration, and usually relies on Debian packages to
ensure stability of the system. But the effort to maintain a stable
system conflicts with installing and maintaining Weblate, a
Python web application, which requires using up-to-date versions of
Weblate itself and of its dependencies.
Having that in mind, and taking into account that we already started
using Docker to replicate the translation server environment to
experiment with upgrading and running an up-to-date version of Weblate,
it can be a good trade-off to use Puppet to provide an environment to
run Docker, and to use a Docker container to actually run an up-to-date
From the present state of the Docker image, which currently uses
(slightly modified/updated) Puppet code to configure the environment and
then sets up Weblate, the following steps could be taken to achieve a
new service configuration as described above:
* Move the database to a separate Docker service.
* Remove all Puppet code from the Docker image: inherit from the
simplest possible Docker image and setup a Weblate Docker image with
all needed dependencies.
* Modify the Puppet code to account for setting up an environment that
has Docker installed and that runs the Weblate Docker image.
* Set up persistence for the Weblate git repository and configuration.
* Set up persistence and backups for the database service.
* Update the Puppet code to run tmserver (if/when it's needed -- latest
Weblate accounts for basic suggestions using its own database).
After that, we should have a clear separation between stable
infrastructure maintenance using Debian+Puppet in one side and
up-to-date Weblate application deployment using Docker in the other
side. The Docker image would have to be constantly maintained to account
for Weblate upgrades, but that should be easier cleaner than deploying
Weblate directly on the server.