Object Storage Model: has anybody real-world experience?

[Stefan Marr]

Hi Raffaello:

> On 16 Mar 2016, at 12:30, Raffaello Giulietti <raffaello.giulietti at supsi.ch> wrote:
> 
> I’m wondering if it has been used even in such class-based dynamic languages like Smalltalk and Newspeak, where objects have a fixed size of (named) dynamically typed instance variables but where the full flexibility of the model might perhaps be an overkill.

Yes, I used it for SOM Smalltalk: https://github.com/smarr/TruffleSOM/tree/truffle-object-model/
It can have benefits there as well.

Imagine a very specific use case, a simply linked list implemented in the standard library.
The class that implements the elements of such a list is typically used to store a wide range of different types.
So, if you decouple the object model from a class, you could allow for specialization of that list element for integers as well as doubles.

This could be a benefit. However, I am not sure how common that actually would be.

In SOMns, i.e., Newspeak, I did not implement the Truffle object model yet, because I got a very similar custom object model (still from my SOM Smalltalk), and in Newspeak, I actually need another version of it for not objects, but classes, because of the inner class semantics.

> Also, I’m wondering if the OSM has been used and measured in real-world server applications with hundred of millions or even billions of live objects and/or if there are informal reports on its the usage in such harsher environments.

That’s perhaps something for the other Truffle languages, I don’t have numbers for such applications.

Best regards
Stefan


-- 
Stefan Marr
Johannes Kepler Universität Linz
http://stefan-marr.de/research/