MetaKit for python (Mk4py) on Macintosh ... no way :(

Gordon McMillan gmcm at hypernet.com
Mon Feb 28 22:40:56 CET 2000


Arnaud Fontaine writes:

[Metakit]

> For me, it's "not really the whole model" ... but as nothing beats the 
> relational algebra at managing tons of structured data ...
> (someone said something about the O2 model ?? u kidding ? I was talking 
> about real world ;-))

> I need to do some good old
> "SELECT obj1.a,obj2.c FROM obj1,obj2 WHERE obj1.d=obj2.e AND 
> obj1.f=value"
> 
> Well .... :)

view1 = db.view('view1')
view2 = db.view('view2')
view2.rename(view2.e, 'd')
v = view1.join(v2, view1.d)
v = v.select(f=value)
for row in v:
  print row.a, row.c

> more details ?
> I've tons of data in objects, I need persistence but can't play with the 
> navigational model to search in those data. Any good way to map them to 
> tables and query the tables ?? (I'm too lazy to do it by hand and I'm 
> not sure I'll be able to do it in an efficient way).
> One more thing : at some steps, I really need the data in objects.

A MetaKit row is an object, but not an object of your making. 
It can have properties that are ints, strings, doubles, or other 
views. So while it's not pickle, it's a whole lot easier to map 
object persistence into Metakit than into a conventional 
RDBMS.

> > Do you care about multi-user operation ?
> 
> Kinda ... in fact, I need multi-processes access to my data. Yes it's a 
> problem with MetaKit.
> 
> > What are your scale and performance requirements?
> 
> About 200 000 objects to search in (but not too big objects) and ... 
> speed needed.
> I've also another need managing millions of object but there, I don't 
> care about speed.

Don't think you'll find anything that meets those requirements 
that doesn't consume vast amounts of resources (including 
$$, probably).
 
> > In what way does Gadfly
> 
> Hey ! (/action takes notes about this "Gadfly") drop me an URL, I'll 
> check this !

www.chordate.com

- Gordon




More information about the Python-list mailing list