Hi guys,
Have been reviewing the efficiency of watirgrid's design of late in
terms of performance. One of the issues I see is having to thread in
user land, in order to execute parallel tests.
To give you an idea, this is the code I would have to write at the moment:
require 'rubygems'
require 'watirgrid'
browsers = Watir::Grid.new(:ring_server_port => 12358,
:ring_server_host => '192.168.1.122', :loglevel => Logger::DEBUG)
@browsers = browsers.start(:quantity => 2, :read_all => true)
threads = []
id = 0
@browsers.each do |browser|
threads << Thread.new do
b = browser[2].new_browser
b.goto("http://www.google.com";)
b.text_field(:name, 'q').set("watirgrid")
b.button(:name, "btnI").click
end
end
threads.each {|thread| thread.join}
Basically create a new thread for every browser in the tuplespace and
then send it the commands. I don't believe this would be very
efficient with say 1,000 threads (if going for a large load test of
sorts).
So IPC (via DRB etc) sorts out the requirement of having multiple
processes (watirgrid providers) executing Ruby code, what I need is an
efficient way to coordinate all this on a single watirgrid controller.
Presently I'm using threads, but am open to ideas? Erlang? Anyone have
experience in this area?
Regards,
Tim Koopmans
On Mon, Nov 16, 2009 at 7:39 PM, Tim Koopmans <[email protected]> wrote:
> Hi guys,
>
> In view of my last email about this I've decided to push on with druby
> instead of http as I think the concepts surrounding distributed
> computing are a good fit. This work has progressed since my last, you
> can view the source here: http://github.com/90kts/watirgrid
>
> I haven't been able to contact any of the original authors of dwatir
> or watirgrid so ploughing ahead under my own direction. I've only been
> testing on a local machine with a windows VM, so the next steps are to
> test this approach in anger.
>
> If you'd like to have a look at the source, or play with the gem feel
> free (gemcutter is the primary source)
> gem install gemcutter
> gem tumble
> gem install watirgrid
>
> I've also included some instructions and detailed examples in the
> source as RDoc.
>
> I'm heading for a railscamp this weekend so planning on tinkering with
> watirgrid some more. As you can probably tell from my source, I'm a
> tester-wannabe-developer so any general advice you have on
> refactoring/structure/approach etc would be appreciated. I'll try and
> round up some advice from railscamp.
>
>
> Regards,
> Tim Koopmans
>
>
>
> On Wed, Nov 4, 2009 at 4:21 PM, Tim Koopmans <[email protected]> wrote:
>> Hi guys,
>>
>> I've recently been tinkering with distributed testing across a grid
>> network with Watir. This work is currently hosted on GitHub at
>> http://github.com/90kts/watirgrid
>>
>> I based my initial design off DWatir[1] which uses Rinda to implement
>> a tuple space: a repository of tuples that can be accessed
>> concurrently. I've tried unsuccessfully to contact the original author
>> Dane Avilla, which looks like development stopped in 2007.
>>
>> Since starting the GitHub project I've noticed a similar solution, not
>> using Rinda (UDP broadcast over druby protocol) but WebBrick (http
>> protocol) instead. It's also called WatirGrid[2] (oops!) and the
>> author looks to be Peter Houghton.
>>
>> My end state is to facilitate some form of load testing with watir,
>> using many browsers concurrently. I plan to release it as a separate
>> gem. My preference so far is to continue with the druby protocol.
>>
>> Questions to you are do you have any preference on choice of protocol
>> (druby vs. http) and does anyone know the original authors so that I
>> might get in contact with them?
>>
>> [1] http://dwatir.rubyforge.org/
>> [2] http://www.ohloh.net/p/watirgrid
>>
>> Regards,
>> Tim Koopmans
>>
>
_______________________________________________
Wtr-development mailing list
[email protected]
http://rubyforge.org/mailman/listinfo/wtr-development
