Removing fuzzy string match in future l10n PR code pushes

[peiying2]

Background:

The l10n team is working on rolling out a pre-translation feature on Pontoon, which is done through deploying machine translated (MT) content for localizers to review before publishing (save) them.

This change will conflict with the current Fuzzy matched strings in Pontoon as they mean different things in this new concept.

More on this: https://github.com/mozilla/pontoon/blob/master/specs/0107-pretranslated-status.md.
What will happen:

The strings currently shown as fuzzy matches in Pontoon will be shown as missing strings, meaning, pending translation. And once the new strings landed, they will go through the MT step and the strings will appear as Fuzzy Matches.

What do devs need to do for future code updates:

For future PRs for l10n, please change to no-fuzzy-match approach as detailed in this doc for gettext file type.

[peiying2]

@akatsoulas @LeoMcA this is a followup from the email sent in case you have questions. You can reach out to flod or matjaž for clarification. Thanks

[akatsoulas]

@peiying2 Sumo uses puente to extract and merge strings which does not support this option (along with many others). We will need to either update puente or switch sumo to a different process for extracting and merging strings. I will open an issue to explore our options and track this effort. What's the urgency here?

[mathjazz]

Hey @akatsoulas! :)

There is no rush. Support for PO Fuzzy strings remains in Pontoon.

The only thing that changes is the visibility of Fuzzy strings in Pontoon. Rather than being assigned a dedicated translation status (Fuzzy), they will now be treated as Missing, which means they will not be accessible from dashboards directly.

While developing this presentational change in Pontoon, an idea has been proposed to stop using Fuzzy strings in the projects completely. Anecdotally, Fuzzy translations are rarely useful and need to be rejected by localizers most of the time.