Is symbol type compressed or indexed in Kdb+?

The official doc already showed that symbol is an atomic, but does this feature utilized for compressing or indexing ?

What we can know is that, in QuestDB, the symbol type is compressed and indexed, "Reduced complexity of database schemas by removing the need for explicit additional tables or joins", but what about in Kdb+ ?

Symbols, which are int mappings to distinct tokens (characters/strings), in q share many of the same properties you've listed from QuestDB, including indexing and compression.

Here is an indexing example:

q)words:`the`cat`in`the`hat
q)words 1
`cat

For compression, as an example check out parted attributes and this whitepaper "Working with sym files"

https://code.kx.com/q/basics/syscmds/#w-workspace

In kdb+ memory there is only one copy of each symbol (a unique char string).
Once you declare a symbol it is internalized (hashed) into the symbol storage structure.

When you work with symbols, kdb+ is using the storage address of the string not the data itself. This reduces memory usage and speeds up many operations.

https://code.kx.com/q/wp/symfiles/

On disk a sym file is used in a similar way. The index in to this file is used in tables column for speed and space efficiency.

In Kdb+, symbols are typically compressed rather than indexed.

Symbols in Kdb+ are a data type used to represent enumerated values or categorical variables. They are stored as a list of unique strings, with each unique string assigned a unique integer index. When a symbol column is created in a Kdb+ table, the unique strings are stored in a symbol table, and the column itself stores the corresponding integer indices.

By default, Kdb+ uses a technique called symbol compression to minimize the memory footprint of symbol columns. The compression is achieved by storing the unique strings in a dictionary-like structure, where each string is assigned a unique integer code. The symbol column then stores the integer codes instead of the actual strings, reducing memory consumption.

When working with compressed symbols in Kdb+, the compression and decompression of symbols are transparently handled by the system, allowing efficient storage and retrieval of symbol data while maintaining the original string representations.

It's worth noting that while symbol compression is the typical approach in Kdb+, it is possible to disable compression and store symbols as simple indexed integers if desired. However, this is less common and may not be the default behavior in most Kdb+ installations.