Re: [myhdl-list] Global state and multiple instances

[Henry G. <he...@ca...>]

On 06/05/15 09:18, Henry Gomersall wrote:
> In light of a PR I submitted last week, I have an idea to try... I'll
> try to get some code later on...

I've implemented a partial solution...

https://github.com/hgomersall/myhdl/tree/globals_free_sim

It's fairly clean and shouldn't break _too_ much. Most of the changes 
are in _simulator.py.

It works by creating a context which can be used to store all the 
current simulation variables.

So whenever access to the simulation variables is required, the context 
manager is used:

with _simulator.simulation_context(some_hashable):
     do the simulation stuff here

There is a default context which it falls back to when not inside the 
context manager.

All the core tests pass except two, in the tracing stuff. The issue here 
is that the simulation now uses its own context which the tracing code 
doesn't know about, so the variables are not shared. This is just an API 
issue so I haven't tried to fix it (i.e. what are peoples opinion, 
assuming the general approach is good).

The other problem is the state on a Signal object is not being handled 
properly. I tried to implement the solution for this (currently 
commented out in _simulator.py) in which on each context switch, the 
signals are stashed (using _Signal._stash_state) to the context that is 
being departed, and then the signal state is reloaded when it is 
reentered (susing _Signal._recall_state). The tests begin to work and 
then one gets into an infinite loop because some waiter or other is 
missing. I'm not sure quite what I've missed. This is hard to bug 
isolate because the tests all run in isolation or indeed, in their 
respective test classes.

Also, the signal code really hits performance, so it needs a bit of a 
reassessment from that perspective too.

I'd appreciate feedback on this approach.

Cheers,

Henry