how to use contains / containsany functions in filter ?

[kran]

I have field named stages like this:

[
  {
    "t": "10262",
    "d": 16593,
    "u": "u1",
    "fs": "A",
    "f": "10262",
    "ts": "A"
  },
  {
    "t": "10622",
    "d": 16593,
    "u": "u2",
    "fs": "A",
    "f": "10262",
    "ts": "B"
  },
  {
    "t": "10001",
    "d": 16593,
    "u": "u2",
    "fs": "B",
    "f": "10622",
    "ts": "C"
  }
]

then I want to fetch records stages have t = 10001, query wrote like this, but not work, always return empty result:

select from SomeType where stages contains(t = '10001')

then rewrote to, and this time got the matched records:

select from SomeType where stages.t contains '10001'

I want to know how contains* functions work

[kran]

and other question is how to remove a column from types ?

[lvca]

You can use the DROP PROPERTY SQL command or Java API.

[kran]

drop property does not work, after drop property, column with null value just returned.
I want to drop column completely (something like mysql).

[lvca]

The contains operator is supposed to work exactly as you described. I wrote a test case and works:

  @Test
  public void testContains() {
    String className = "testContains";

    DocumentType clazz1 = database.getSchema().createDocumentType(className);
    clazz1.createProperty("list", Type.LIST);

    database.getSchema().createDocumentType("embeddedList");

    database.transaction(() -> {
      for (int i = 0; i < 100; i++) {
        MutableDocument document = database.newDocument(className);
        document.set("list", new ArrayList<>());

        for (int j = i; j < i + 3; j++)
          document.newEmbeddedDocument("embeddedList", "list").set("value", j);

        document.save();
      }
    });

    int totalFound = 0;
    for (ResultSet result = database.query("sql", "select from " + className + " where list contains ( value = 3 )"); result.hasNext(); ) {
      Result item = result.next();
      List<EmbeddedDocument> embeddedList = item.getProperty("list");

      List<Integer> valueMatches = new ArrayList<>();
      for (EmbeddedDocument d : embeddedList)
        valueMatches.add(d.getInteger("value"));

      Assertions.assertTrue(valueMatches.contains(3));

      ++totalFound;
    }

    Assertions.assertEquals(3, totalFound);
  }

[lvca]

(Commit 5e7dfb7)

[kran]

thanks @lvca.

i'll try it in this weekend.

but here i have a question, why it's a map?

stages.type() gives list as the answer.

[lvca]

Yes, it's a list of maps instead of a list of embedded documents. Maps and embedded documents are very similar. The reason must be in using the merge from a json. The JSON is converted in map(s).

[kran]

ok, got it, i will try again.

luca, i like the idea of arcadedb, like the project a lot. and using it in company products. but with the experience of these weeks, i have to ask myself if it's a right choice. i can bypass bugs in sql layer, but then i'm a little worried about the stablity of storage engine layer.

i always know that it is my responsibility not yours. but i just worry about it.

forgive me if i'm not so friendly.

thank you.

[lvca]

@kran that's a legitimate question! :-) The ArcadeDB engine is 100X more stable than OrientDB which has a much longer history. Unfortunately, we forked the entire SQL engine from OrientDB inheriting all its bugs. While nobody is fixing those bugs in OrientDB anymore, we're active on ArcadeDB. I have to say that the most stable part of ArcadeDB is the engine, and other layers (like the Postgres protocol) are not very mature and there is still a lot of work.