Does WinRunner offer any type of Multi Client type testing?
User Misha (email@example.com) posted:
For this sort of test, there may be several ways to implement with
WinRunner, which doesn't have anything built in for this unless you count
its interaction with TestDirector.
If you can log in as more than one client session on the same PC at the same
time, WR could swap between windows of the client, and the test
synchronisation is then straight forward.
If you need to run the different clients from separate PCs, running WR on
each one, you will need a way for the tests to send messages to each other.
The simplest way to do this is writing to files in a shared directory, for
instance the Leader test you mentioned may want to poll the directory for
any Participant messages. Unless you are writing to a shared database, I'd
advise not to try writing to a single file from different PCs, the file
locking is hard to implement in WR, its easier to stick to a separate file
for each Participant.
There are several other options, including setting up a message queue, which
is easiest depends on what technologies you have to hand.
I found it rather fun to watch one test waiting for the results of another
to continue, and influencing its behaviour.
From: "mewongster" <mewongster@h...>
Date: Mon Jun 24, 2002 2:42 pm
Subject: Does WinRunner offer any type of Multi Client type testing?
We are trying to set up a Multi Client type test, and I am trying to
figure out if WinRunner ahs the capability to automate.
For example, a Participant within a online web meeting session
selects a YES type functionality button within their client
application, executing a check mark next to their name, and the
Leader conducting the session is able to view the YES (check)
responses on his own client application.
Has Mercury worked with a Multi Client type test before? Where a
Leader can view the interactions of the many participants, and the
Participants can interact with what the Leader requests?
Re: Does WinRunner offer any type of Multi Client type testing?
Just an added note on coordinating your machines. If you are running, for example, under NT, you could have your scripts use command prompt based "NET SEND" commands. The scripts would have no problem waiting for a "Messenger Service" window and checking its contents.
This should be fun to implement.