HDFql working with tables

I have some question on the usability of HDFql:

  1. Does HDFql support created of a table where the row are of different type?
  2. How do you append data to the table?
  3. How do you iterate though the row do the table?

The table I want to create will have any thing 1 to 2^n rows and cannot be determine before hand.


  • Here goes some info concerning your questions about HDFql:

    1. If by "HDFql support created of a table where the row are of different type" you mean HDFql supports compound data type, the answer is not yet. (EDIT: since HDFql version 2.2.0 compound data types are now supported)

    2. To append data into a dataset (where its size cannot be determined before hand), you have to go through several steps (I will assume that you are using the C programming language):

      2.1. The dataset must be extendible. As an example, you can create an extendible dataset in HDFql as follows (this creates a dataset named dset of integer data type with an unlimited size):

        hdfql_execute("CREATE CHUNKED DATASET dset AS INT(UNLIMITED)");

    2.2. Write a value in the last row of dataset dset using a hyperslab as follows (replace my_value with an integer that you would like to write into the dataset):

        hdfql_execute("INSERT INTO dset[-1:1:1:1] VALUES(my_value)");

    2.3. After writing a value to dataset dset and if there are more values to write, first increase (i.e. alter) the dimension one unit like the following and then repeat step 2.2.:

        hdfql_execute("ALTER DIMENSION dset TO +1");
    1. To iterate through the rows of dataset dset, you must first read it and then use function hdfql_cursor_get_int() as follows:
        hdfql_execute("SELECT FROM dset");
        while(hdfql_cursor_next(NULL) == HDFQL_SUCCESS)
            printf("Value: %d\n", *hdfql_cursor_get_int(NULL));