Is This by Design or...?

[tetotechy]

I have lost some time figuring out this one. When referencing a table record in a query, SurrealQL treats strings with patterns like identifier:identifier as pointers. So there is no difference between select * from Mouse:Mickey and select * from "Mouse:Mickey" (unless you end up duplicating records, see below).

This is for me the first confusing point, because I would expect select * from "string" to always return the string, independently of the string content. But if I try select * from "useless:string" I get a "result": [], "status": "OK", ... kind of answer instead of "result": [ "useless:string" ], ... as for any other string with a different pattern

Secondly, as I hinted above, because of an inconsistent and likely inappropriate use of js-driver functions (db.create) and literal queries (db.query) in my code I found myself duplicating records in a table, some having the Mouse:Mickey id and others the "Mouse:Mickey" id. This was hard to spot, because the json outcome from queries is by definition all stringified and ids appear as identical strings in both cases (i.e., no double double-quotes for "Mickey:Mouse" or the like)

Congratulations on your amazing work! And apologies in advance if I am missing some relevant point, just started digging into your fantastic product!

[Amareis]

I really think what in strict mode all ids (and maybe even just tables names) should starts with colon - so it always clear that is record link. :city:london is easy to write, read, don't confusing with strings and have some roots with symbols from ruby.

And even in json output, such format much easier to immediately identify ids. And it machine-friendly too - just checks first string symbol, instead of full patter matching.

Additionally, you can read about better type safety at #161

[tetotechy]

after further digging I found there was a bug with numerical ids (my example was not quite correct, actually. duplication of records was triggered by RecordIDS like person:10). apparently this has been solved and will be patched soon (see #149).

still, I think literal strings should be returned verbatim independently of content in cases like select * from "table:id" and hopefully the patch will address that too