[mext.reaction] Why were the "collections" removed?

[Reguillo]

https://bitbucket.org/mlk/reaction/changeset/a9ae7f12d31b

Are they going to be replaced in the future or do we have to write our
own classes for them?

[Sergey Schetinin]

Well, basically it's because they depend on @todo and todo breaks
certain properties (single writer per cell per txn) that are required
to implement per-rule rollback.

I did this and and a couple other changes with the intention to
finally implement a lot of things I had planned for Reaction, but
shortly changed my mind about it. My current stance on Trellis /
Reaction is that it's a great approach that people should be more
aware of and that the tradeoffs it provides are great for a lot of
things, however, to really make the best use of it, support from the
runtime is necessary. Reaction as it is is pretty much at the limit of
what is feasible to tack on top of Python runtime without a deeper
support for certain transaction primitives and without making some
invariants inherent to the language or runtime itself.

In particular new cell / component initialization from inside the
transaction is a huge problem to solve with the tools available and
without it there's a limit on the scale of the project /
'cell-network' Reaction can handle. So I reluctantly have to admit
that this is the end of the road for this incarnation of Reaction.
Maybe I'll eventually give it another go, but it will be a fresh start
and probably as some kind of experimental language extension or a
completely new runtime implementation (I still love Python syntax, so
I wouldn't stray too far from it).

So, if you want to use collections, just use the revision that still
has it. I guess I should mark the revision before some of the code
removals as a new version and publish it on pypi, so it is easier to
depend on.

I most likely will not do any more work on this version of Reaction,
so if anyone wants to take over, let me know (I doubt anyone does
though).

-Sergey

> --
> Mailing list: http://groups.google.com/group/better-python
> Unsubscribe: better-pytho...@googlegroups.com

--
Best Regards,
Sergey Schetinin

http://self.maluke.com/ -- My articles and open-source stuff
http://www.maluke.com/ -- My commercial software and custom development services

[Reguillo]

Thanks for your reaction.
I'm a big user of Reaction and like it very much. Until now I didn't
see any major problems with it.
In my code I use new cell / component initialization from inside the
transaction. I don't know if we are talking about the same thing here.
But so far I didn't see any problems in my tests with it.
I would like to provide some code if anyone is interested.

Thanks for all the work you've done with Reaction.
I don't know if I know enough about it to maintain Reaction.
But i'm willing to learn.

Best Regards,
Reguillo

> On 29 December 2010 16:36, Reguillo <reguil...@gmail.com> wrote:
>
>
>
> >https://bitbucket.org/mlk/reaction/changeset/a9ae7f12d31b
>
> > Are they going to be replaced in the future or do we have to write our
> > own classes for them?
>
> > --
> > Mailing list:http://groups.google.com/group/better-python
> > Unsubscribe: better-pytho...@googlegroups.com
>
> --
> Best Regards,
> Sergey Schetinin
>

> http://self.maluke.com/-- My articles and open-source stuffhttp://www.maluke.com/-- My commercial software and custom development services