More 'About Storm'

[Michael Bayer]

On Jul 12, 2007, at 10:25 AM, Mark Ramm wrote:

> Thanks for the quick response.   Fundamentally the problem is your
> opening line to the list which claims that those are the things which
> make Storm different from other projects -- and I don't think that's
> what the list is for.
>
> To be more specific, unless I'm reading your list wrong SQLAlchemy
> does have all the "features" on your list.
>

no, our native two-phase commit support is still a patch in trac,  
waiting to be cleaned up and implemented, and theres also some talk  
that we should get xa-transactions in there as well.   of course  
theres no architectural issue that prevents SQLAlchemy from having  
this feature.

otherwise yes the whole list, so far, has nothing unique to Storm  
(including test-driven development..we are up to 749 unit tests right  
now).   Looking over the tutorial there are various features and  
behaviors of the Store which we can also implement if people express  
a desire for them...also i havent yet seen a whole host of features  
SQLAlchemy has, such as ad-hoc eager loading, all three kinds of  
class inheritance, polymorphic loading, cyclical foreign-key  
relationships, self-referential relationships, mapping to selects and  
joins, custom types, composite types, etc., but im sure some or all  
of these will be revealed soon.     im also looking forward to seeing  
the details of how Storm does object/row-level multidatabase access,  
which is another feature which we do not yet have  
implemented....we'll be sure to evaluate its design for clues towards  
our own approach.

i think the differences, as Gustavo mentioned, will come down to  
stylistic choices and individual case histories.  you can post a lot  
of brags on your site about "we're the ONLY ORM TO DO THIS", but more  
sophisticated users will see right through it (and Python is, after  
all, a sophisticated set of users. its why I chose this language).