Re: Implicit Wait

On 17/10/16 22:59, Luke Inman-Semerau wrote:
> I like Jason's proposal much better. Client side maintaining it (we can
> use some explicit waits under the hood for the user).

I think that implicit waits are a bad idea. But before removing this 
from the protocol please consider if this change is worthwhile given the 
timetable for committing to protocol stability (~4 months). I think the 
same criteria as the value wrapper changes should apply i.e. explicit 
buy in from major implementors.

With my geckodriver hat on I am concerned that there will be an 
indefinite period where we will implement the hypothetical no-wait spec 
and others will implement implicit waits and users will be convinced 
that geckodriver is unreliable and broken.

(For related reasons, and on a different topic, I am quite concerned 
about the changes to new session handling. I think the PR is a win in 
the abstract, but I am concerned at the size of the breaking change vs 
the apparent preference of other implementors to prefer compatibility 
with shipping selenium vs targeting the spec).

Received on Monday, 17 October 2016 22:14:25 UTC