When searching for a string that is present on the meta in a url (e.g. discourse-rewards) there are no results coming up even though this string exists in a url posted on meta. I’m not sure if this has bene discussed previously, I apologize beforehand in case my question is redundant.
סיכום מבוסס בינה מלאכותית
The discussion addresses a bug where the search function fails to find keywords in URLs, even when they exist in links posted on Meta. Mr.X_Mr.X reports that searching for a string like ‘discourse-rewards’ in a URL does not return results, despite its presence. sam explains that the issue stems from how URLs are tokenized, noting that the system now limits URL splitting to avoid index bloat, resulting in partial URL fragments like '/ahmedgagan/discourse-rewards' being indexed. While this change was intentional, sam acknowledges it could be refined, though it’s not a high priority. Mr.X_Mr.X agrees, noting this is a common issue when searching for plugins shared as links, suggesting it could be improved in the future.
2 לייקים
I guess you are talking about:
The post tokenizes to:
'/ahmedgagan/discourse-rewards':23 'admin':34 'alreadi':8 'brows':30 'com':22 'contribut':3A 'discours':39 'featur':11 'find':17 'forum':2A 'github.com':22 'github.com/ahmedgagan/discourse-rewards':21 'instal':26 'marketplac':6B 'paid':1A 'plugin':14,19,28 'point':4A 'reward':40 'search':37 'set':32,35 'system':5A
I recall @tgxworld limited some of the URL splitting we do:
so: https://github.com/Ahmedgagan/discourse-rewards
Is now being split to: github.com and github.com/ahmedgagan/discourse-rewards and '/ahmedgagan/discourse-rewards'
I think limited splitting on path was deliberate here so we don’t bloat the index too much. We may be able to refine this, but not in the urgent list.
4 לייקים
I often encounter this on meta when searching for a plugin that is posted as link. I guess in the future it could be a good idea.
3 לייקים