lint rule: "a message should only be used once"

[samuelstroschein]

Problem

Referencing the same message in multiple places leads to a broken message.

No guarantee exists that a message with the content like "learn more" won't be changed by someone to "learn more about unlimited inboxes". If the message is rendering all "learn more" references, the example website below will break at 2 and 3.

Proposal

A lint rule that detects if a message id is used multiple times in source code.

// unlimited inboxes
<p>{m.elephant_big_tree()}</p>

// manage team members
// 💥 error m.elephant is used multiple times
<p>{m.elephant_big_tree()}</p>

// spam report
// 💥 error m.elephant is used multiple times
<p>{m.blue_box({ title: m.elephant_big_tree() })}</p>

(Creating new messages is cheaper than trying to manage shared messages.)

Additional information

• emerged in a conversation on discord https://discord.com/channels/897438559458430986/1206732202839253052/1208848809304530974
• requires plugin API where is a message used in source code for "unused message", "hardcoded text", etc. lint rules #2126

[samuelstroschein]

cc @martin-lysk who is a strong advocate of a message should only be once

[jannesblobel]

I'm curious about the pros and cons from @martin-lysk's perspective. :)

[martin-lysk]

@jannesblobel: @samuelstroschein outlined the most important points here. The example of "learn more" that should be changed to "learn more about xyz" in the middle of a project is the beginning of a lot pain.

"One message per context" as a rule of thumb will usually solve that. While this means more translation work the overhead is worth it. Most of the TMS out there introduce features like "translation memory" – saving a reviewed translation for a given term to reduce the amount of translation work - something we should keep in mind as well.

[samuelstroschein]

discussion moved to opral/inlang-sdk#7