PyTable?

Discussion in 'Python' started by Chris McAvoy, Mar 1, 2004.

  1. Chris McAvoy

    Chris McAvoy Guest

    Hi,

    I was looking at PyTable this morning, after seeing it on PythonWare's
    Daily Links page. It's interesting, because my local Perl Monger's
    group is having a talk on Class::DBI tomorrow night, and they sound
    very similar. I can't attend the talk tomorrow, so thought I'd ask
    something here.

    What are these two modules about? Another abstraction layer for the
    DB? I've never heard of this as a concept until today, and I'm a
    little lost. Are they mimicing functionality available in another
    language that I'm not aware of?

    PyTable points to another module called BasicProperty that abstracts a
    Class?

    Could someone help me wrap my head around what appears to be a new
    "trend"?

    Thanks,
    Chris
    Chris McAvoy, Mar 1, 2004
    #1
    1. Advertising

  2. Chris McAvoy wrote:
    ....

    >What are these two modules about? Another abstraction layer for the
    >DB? I've never heard of this as a concept until today, and I'm a
    >little lost. Are they mimicing functionality available in another
    >language that I'm not aware of?
    >
    >PyTable points to another module called BasicProperty that abstracts a
    >Class?
    >
    >Could someone help me wrap my head around what appears to be a new
    >"trend"?
    >
    >

    Will try, though since I authored the sites that confused you I may not
    be the best person to try. First of all, I can't really say anything
    about the Perl module, as I don't use Perl. I'll just stick to what
    PyTable does.

    PyTable's primary strength is its schema-driven nature. A database plan
    (schema) is created which maps the structure of the database into a
    run-time introspectable data structure.

    Normally the schema is created in Python code, and used to drive the
    creation of the database, (though facilities are available for
    extracting a basic schema from the database). The schema can include
    information regarding what classes to instantiate when retrieving
    records for a given table (with suitable defaults provided). The schema
    is used internally by DBRow objects to implement update and insert
    queries, for instance. In other words, the DBRow objects can look at
    themselves to decide what actually needs to be done to accomplish a
    given task (such as specifying their row uniquely in the table).

    PyTable sits halfway between the raw DB-API (where you know only
    approximate data-type, name, and a few other minimal pieces of
    information) and a system such as PDO (where the database interactions
    are done through an object-oriented API). PyTable is still very much
    about generating and running SQL code, you are not forced into the
    "everything is an object" mould, you can still run raw SQL (with some
    helper mechanisms to make creating complex queries easier) and integrate
    those queries into your row classes. The PyTable DBRow objects are
    simply very rich introspectable wrappers around the DB-API records (you
    can use non-DBRow records too).

    As to BasicProperty, its purpose is to allow for defining rich
    "attribute descriptors", similar to the property objects built in to
    Python 2.2 and above. Distinction being that the BasicProperty
    descriptors are far more involved, including default values/functions,
    two levels of storage hooks for customizing behaviour, the ability to
    auto-coerce values to/from data-types on save/restore, support for
    list-of types etceteras. BasicProperty is an extremely general
    mechanism, so it's a little hard to say what particular benefit it has
    without looking at particular applications.

    * In PyTable, the field descriptors are actually BasicProperty
    objects linked to the field schema for the table from which they
    were created, so they can, for instance, be used to lookup foreign
    keys (and their tables)
    * In PyTable's internal machinery, just about everything is a
    propertied object. This allows for defining objects as a
    collection of properties with documentation, default-values,
    data-typing and coercion. There are very few "__init__" methods,
    as just about everything uses the same base class which provides
    simple property-aware initialisation. Adding/removing properties
    generally doesn't require any change to the code other than the
    actual property declaration.

    Back to the subject at hand; In effect, PyTable is about trying to make
    the nitty-gritty details of dealing with an SQL-driven application
    easier by providing introspection facilities that let your code reason
    about the structure of the database (as embodied in the schema objects).

    What PyTable is not:

    * Lightweight: compared to db_row or the like, it's very
    heavyweight, focused on creating interactive apps, not huge
    back-end processing apps that need to realise millions of rows
    simultaneously
    * An SQL abstraction mechanism: unlike PyDO no attempt is made to
    define a common SQL sub-set, i.e. it's not like ADO from Microsoft
    * An object-relational mapper: though it may seem like it, PyTable
    doesn't convert references to records automatically (unless you
    explicitly code a property to do so), it doesn't map pivot tables
    to pseudo-dictionaries, in general, it only executes the SQL you
    tell it to execute
    * Finished: there's still a lot of work to do to get to 1.0 beta 1
    for PyTable

    Anyway, hope this helped somewhat. Have fun,
    Mike

    _______________________________________
    Mike C. Fletcher
    Designer, VR Plumber, Coder
    http://members.rogers.com/mcfletch/
    Mike C. Fletcher, Mar 1, 2004
    #2
    1. Advertising

Want to reply to this thread or ask your own question?

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.

Share This Page