(original) (raw)

> Do you have any use case for modifying a variable inside some context?

> numpy, decimal, or some sort of tracing for http requests or async frameworks like asyncio do not need that.

Maybe I misunderstood how contextvars is supposed to be used. So let me give you an example.

I understand that decimal.py will declare its context variable like this:

---

contextvar = contextvars.ContextVar('decimal', default=Context(...))

---

Later if I would like to run an asyncio callback with a different decimal context, I would like to write:

---

cb\_context = contextvars.copy\_context()

decimal\_context = cb\_context\[decimal.contextvar\].copy()

decimal\_context.prec = 100

cb\_context\[decimal.contextvar\] = decimal\_context # <--- HERE

loop.call\_soon(func, context=cb\_context)

----

The overall code would behaves as:

---

with localcontext() as ctx:

ctx.prec = 100

loop.call\_soon(func)

---

I don't know if the two code snippets have exactly the same behaviour.

I don't want to modify func() to run it with a different decimal context. So I would prefer to not have to call decimal.contextvar.set() or decimal.setcontext() in func().

But I would need contextvars.Context\[var\]=value to support such use case.

Decimal contexts are mutable, so modifying directly the decimal context object would impact all contexts which isn't my intent here.

Victor