בדיקה בשפות שאינן לטיניות

מפתח

סיכום מבוסס בינה מלאכותית

A discussion on testing Discourse with non-latin languages is underway. chrispanag shares their experience with a Greek forum, highlighting several bugs related to non-latin characters. They wonder if automated test cases using non-latin languages could prevent such issues and found no existing tests for this purpose. Falco invites chrispanag to create and submit a pull request for such tests, which chrispanag is willing to attempt. They request guidance on where to start, specifically asking if they should extend existing fabricators and seeking an example to get started.

1

After 3 years of using Discourse for a greek forum, I have identified various little bugs that go unnoticed between updates and have to do mostly with non-latin characters.

For example:

  1. Some categories are broken: "Load cannot follow more than 20 redirections"
  2. Unicode username with Σ as the final char leads to an error loading profile page
  3. Topic with Japanese in URL doesn't redirect if URL doesn't perfectly match

Although some of those could just be edge-cases, those issues have left me wondering if they could be preventable by implementing some automated test cases using non-latin languages.

I did a quick browsing at the various tests that discourse is shipping with and I didn’t found any any that are testing against non-latin fabricated data.

My questions are:

  1. Are there any such tests that I missed?
  2. If not, wouldn’t be helpful to add such tests so that some of those bugs could be catched before deployment?
2

If you can add such tests and send a pull-request that will be welcome :wink:

3

I could try to do that :slight_smile:

Do you have some pointers on where I should start? Should I extend the already existing fabricators and if yes, can you share an example? From that, I think I can continue and add some more tests.