#5025 - Searching title/meta tags in a Gallery category fails to populate results
| Identifier | #5025 |
|---|---|
| Issue type | Trivial issue (does not break functionality) |
| Title | Searching title/meta tags in a Gallery category fails to populate results |
| Status | Completed |
| Handling member | Chris Graham |
| Version | 10.0.43 |
| Addon | core_database_drivers |
| Description | Even following the links from the meta tags which match the title yields nothing. |
| Steps to reproduce | |
| Funded? | No |
| Commits |
The system will post a comment when this issue is modified (e.g., status changes). To be notified of this, click "Enable comment notifications".


Comments
Meta tags on the category such as Canada, Female Fronted Metal, Melodic Death Metal return results
Adding Your Last Wish as a meta tag weirdly also fails to return anything (but I'd prefer a return on the title than adding an additional tag for the band name anyway).
Other category titles are returning results in search, so maybe this is stop words or something?
(Click to enlarge)
https://dev.mysql.com/doc/refman/8.0/en/fulltext-stopwords.html
MySQL's fulltext stoplist includes the word "wish" and possibly other reasonably distinctive words, which can break some perfectly reasonable queries working.
We already have support for switching out a stopword search to a simple string comparison. Extend that to detect a compound-stopword search. Also allow it to do a prefix-locked LIKE query if the search is at least 8 characters; which will be indexable and reasonably distinct, so not too much of a performance penalty.
Likely MySQL found those words by statistical analysis of common words in text. The idea is that indexing wastes space and searching wastes time and relevancy suffers unless common words are first taken out of the equation.
1) Performance, as now any short search matching stopwords becomes a LIKE query which is more intense. It's a possible DOS-attack vector. This is all based on my perceptions, every site will have different performance based on the nature and quantity of content, and server resources,
2) You presumably would only want "Isn't" to match (which will work with my change), but not "Isn't It Wonderful" or whatever. The character limit of 8 was also intended as a tradeoff to resolve this problem.