We're Hiring!

Mattermost, Inc.

Searching 1-letter and 2-letter phrases

We have a stocks-trading community that migrated from Rocket.chat to MM E0 self-hosted. Many stocks are one or two characters long (e.g. $AA, $BB, $U). MM limits search to 3-digits or greater. This makes it super difficult to find key messages around these.

From what I can gather, this limitation may be part of MM core… I REEALLY dont want to modify core if I can help it but this is a major problem. We NEED to find mentions of certain tickers. .

Before I consider modifying the core codebase to accommodate I wanted to confirm that im not missing a setting somewhere first…

Also helpful would be knowing where in the core this setting is located in core (were I forced to go that route)

Can I ask if you are using the elasticsearch search indexing features? I (from my use of ElasticSearch in the ELK stack, and other cybersecurity-related uses) was under the impression that this was possible through their engine. Is this something that the Mattermost code force limits then?

Yeah, we’re using Elasticsearch.

Hi, @bsegovia

By default, Mattermost does not support this based on documentation:

Like many search engines, common words such as the , which , and are (known as “stop words”), as well as two-letter and one-letter search terms, are not shown in search because they typically return too many results.

Specific to this one:

May I know what articles / documentations that you came across mentioning about this being a limitation of Mattermost core? I would like to double confirm this with the engineering team if that is the real case and how we can go around this (I think Elasticsearch and Bleve still has this limitation too).

Hmm. Well I need to find a way to return 1 and 2 letter results. Our specific context requires it. (Our members are stock traders who search for charter about symbols. Several of which are 2 and 1 characters long)