South Florida Code Camp 2011

Presenting at the South Florida Code Camp 2011 today.  Similar presentation to past code camps and hoping to get a similar turn out.

Slides

An Introduction to UI testing

SWAT 3.0 Released

We are very excited to announce the release of SWAT 3.0.  The biggest announcements are the new support of Chrome on the Windows OS and Safari on OSX.  In addition we now support calling SWAT from the command line independent of any test runner tool.   Finally we have made a roughly 30% improvement on the speed of SWAT running against the IE browser.  As usual we have also fixed a number of bugs and smaller issues that are detailed in the complete release notes.

New Features and Updates

1. New Browsers Supported
   1. Chrome
      1. New custom built plugin for Chrome allows SWAT to easily control the Chrome Browser
      2. http://ulti-swat.wikispaces.com/Setting+up+Chrome
   2. Safari
      1. Custom app for OSX now allows SWAT to stream commands to an OSX based system.
         1. http://ulti-swat.wikispaces.com/Setting+Up+Mac
         2. http://ulti-swat.wikispaces.com/Setting+up+Safari
2. SWAT Console
   1. Tests can now be run independent of any test runner.
   2. As of now this feature is for SWAT only and does not support any custom fitnesse fixtures.
   3. http://ulti-swat.wikispaces.com/SWAT+Console+User+Information
3. New Commands
   1. UI
      1. AssertJSDialog
         1. Added an optional timeout to allow users to select how long the command will search for a dialog box
      2. AssertTopWindow
         1. Allows users to verify if a window is the top most window on the screen
         2. http://ulti-swat.wikispaces.com/AssertTopWindow
   2. Database
      1. RestoreAllTables
         1. Users can now restore all of the tables they have saved at once instead of one at a time
4. Defects Resolved
   1. CORE – AttachToBrowser always finds the about:Blank window
   2. CORE – DB Connect now automatically closes the old connection when it creates new ones
   3. CORE – FF – AssertElementExists will find an element when the identifier is empty
   4. CORE – FF – PressKeys does not ignore CancelEvent in Javascript
   5. CORE – FF – Special characters above the standard 128 return correctly when
   6. CORE – FF – The AttachToWindow failure due to “OK” being added to title has been fixed
   7. CORE – IE – attachToModalDialog is now setting the currentIEBrowser
   8. CORE – IE – Browser screenshots work correctly in IE9
   9. CORE – IE – KillAllOpenBrowsers is now respecting AbandonTest
   10. CORE – IE – OnBeforeUnload JSDialogs in IE9 are compatible with SWAT
   11. CORE – IE – PDF pages open automatically in IE9
   12. CORE – Screenshots don’t throw a “Not attached to window” exception.
   13. CORE – SQL checks are passing in the editor and not in Fitnesse
   14. CORE – SWAT waits for all frames to load when waiting for the browser
   15. EDITOR – Editor is respecting the -c attribute for included files
   16. EDITOR – Values of 0 can now be saved in the Delay between commands option

Burden of Proof

Came across this article today

http://www.androidguys.com/2011/01/01/android-sms-bug-exist/

In it the author comments “But the burden of proof is on those claiming a bug exists.”  Which to me leads to an interesting question.

Does the burden of proof lie on those claiming a bug exists or on the company to prove a bug does NOT exist?

To me the burden lies with the company not the user.  To a user the bug exists, even if the company has not seen it.  As Josh Rynne (@jrynne) pointed out “good business practice would suggest that the company should prove that the bug does not exist…. otherwise you’re basically saying you don’t believe your customers by forcing them to prove it”.  Case in point is the antenna gate scandal where Apple placed the onus on the consumer and not on the company.  It cost quite a bit in bad publicity when Apple had to retract its denials and accept responsibility.

However I believe we can take this idea one step further.  While discussing this with Chris Taylor (@agiletester) his comments were “In reality, in good design, there is no such thing as user error.  There may be dumb users, but you are not going to make money saying it.”  A truly good design should not allow users to make user errors.  Relating back to this article, if the issue really was user error in how they were switching between conversations, the design should have helped users avoid input mistakes in that sort of scenario.

So to answer my own question, the burden of proof for any bug truly lies on the creator not the user.  It is only the users responsibility to alert the company to the issue and the company has to determine if the bug is in the working of the application or the use.

Cloud Based Management

This blog is probably not what you think it will be about.

Where I work we have a very flat management structure.  What this means to us is that their are very few steps between an employee and a director.  In fact virtually all members of a team report directly to a director.  Currently there are about 200 members of the staff and 2 directors (though we are scaling up to 4).  As part of this, we have created what I am calling a cloud based management team.  What this means is that while people have a direct supervisor, they tend to go to any supervisor to discuss any issue.  If director A is busy, people go see B and vice versa.  This tends to allow people the ability to talk to people that they are most comfortable with and also allows more access to our directors since they are so interchangeable.  Overall I have found this to be an incredibly simple, yet effective way to manage a large group of people without a large amount of overhead.  It is definitely not a perfect solution (which I believe to be one of the reasons we are scaling up), however it has been working well for us.  I believe that as we scale up it will be more like adding nodes to the cloud and not forcing people to deal with one specific manager.

SWAT’s International Release

Really enjoyed doing some data mining recently on SWAT’s international reach. Below is some of the download data from 2/1/2010 until 8/13/2010.

1.	United States	615
2.	India	326
3.	United Kingdom	86
4.	Unknown	84
5.	Germany	55
6.	China	50
7.	Korea	49
8.	Canada	46
9.	France	33
10.	Brazil	29
11.	Australia	23
12.	Viet Nam	23
13.	Indonesia	21
14.	Israel	21
15.	Romania	21
16.	Italy	20
17.	Philippines	18
18.	Turkey	16
19.	Netherlands	16
20.	Russia	15

UltiFit goes Open Source

About a year ago I had the good fortune to attend the AAFTT 2009 meeting in
Chicago. While there I demonstrated an internal Windows based tool for running
Fitnesse tests that we had named UltiFit that we were hoping to release to the
open source community. Since then we have added some great new features and
spent over 90 days on stability and reliability testing by supporting our entire
suite through a full release cycle with no issues reported. That being said, I
am extremely excited to announce that it is now available for download at

http://sourceforge.net/projects/ultifit/.

The primary purpose for the development of UltiFit was to create a tool for
“push button regression” of your Fitnesse-based testing inventory, achieving
overnight (for multiple platforms) what might otherwise require an entire team
working all week (for a single platform). When we started using Fitnesse
combined with our UI testing tool (SWAT) we found that tests would fail and
cause a systematic collapse of all of the tests after it. Additionally the
standard fitnesse suite interface did not allow us to run tests from multiple
suites at once. With that in mind we went about creating a more robust test
runner. Initially UltiFit allowed the execution of a single test at a time.
Since then we have added grid functionality which allows a cluster of machines
to share responsibilities for running Fitnesse tests. Internally we use UltiFit
to run our UI automation tool (SWAT) and running just one team’s tests could
take as long as 12 hours. Once the grid functionality was introduced we were
able to drop that time to under 4 hours. Since we always have the option to add
more servers to the grid, we can shave even more time if we choose.

Some of the additional benefits of UltiFit include:

• Hands free automated execution/management of a group of individual Fitnesse/SWAT tests
• Recovering from any issues during execution
• Killing tests that are stuck for some reason
• Organizing/reporting the results and system information
• Archiving history of test results making it easy to find exactly when a test failed along with complete details of what failed
• Detailed timing information captured for each test and for total group of tests
• Easily re-running failed tests (only)
• Command line interface enables scheduled execution
• Grid based execution can dramatically improve regression time (Both XP and Vista supported)
• Loosely coupled, easily configurable design:
• • Involves a single .exe for regular or grid-based deployment
  • Tool is easily configured for any new Fitnesse test inventory

In short, UltiFit allows us to execute and manage our tests while we sleep.
Hopefully at least 1 or 2 others may find this useful as well. If you have any
questions or are interested in helping out with development please let me know
or visit us at http://sourceforge.net/projects/ultifit/support.

Accessing a POP3 Email Account

As part of an internal development project I wanted to be able to access a pop3 email account. After a larger amount of searching then you would think necessary I came across http://www.stardeveloper.com/articles/display.html?article=2009070801&page=1 . After that it was pretty simple. The only major changes I felt were needed were to change some of the logic on how a connection is closed. The source code has all of the email revert to its initial state when the box is done being accessed. However for my needs I wanted to ensure the emails were deleted. To do so I needed to be a little more familar with POP3 commands. I found this to be a good resource for all PPO3 commands (http://www.electrictoolbox.com/article/networking/pop3-commands/).

Mobile Browser Sniffing

For my internal site, I wanted to give it the ability to display an version specific for mobile browsers. To do this required the ability to sniff our a mobile browser. Initially I was using the provided .Net command for verifying if the accessing browser was a mobile browser.

Request.Browser.IsMobileDevice

However while this works great for the iPhone this did not work at all for Android based devices. After a large amount of searching I found the MobileESP open source project (http://code.google.com/p/mobileesp/) for determining if a browser is from a mobile device. Initially I used thier entire code base. However I felt that solution was too bulky and I ended up just finding the code for locating an Android based device and using that instead.

Request.UserAgent.ToLower().IndexOf("android") != -1

Even if you do not need their entire code base, the project is a great repository for determining how to identify any mobile based browser

Creating an iPhone\droid friendly website

For an internal tool I recently created an iPhone\Android friendly version of one of my sites.  Considering the plethora of CSS styling sites you would think at least one or two would have some easy layouts for mobile devices.  Unfortunately this was not the case.  However I did find (http://net.tutsplus.com/tutorials/html-css-techniques/flexible-mobile-first-layouts-with-css3/) which made creating my mobile app super easy.  Creating a mobile site comes down to two things.  Having the Viewport Metadata and having a valid CSS file.  Another useful site (http://eisabainyo.net/weblog/2009/06/12/making-a-website-iphone-friendly-using-css/)

Meta Tag

This tag is needed to help autofit the page and to prevent zooming.  (If you ever wondered why some sites on the iPhone do not zoom, this is the answer)


<meta name="viewport" content="width=device-width; initial-scale=1.0; maximum-scale=1.0; user-scalable=0;" />


CSS

To create my app I copied the layout here

I used essentially the same css as available on the above site.  The most important note I did find in my research however, is not to use fixed widths, and instead to use percentages.

Icon

For websites designed for mobile browsing, it is important to have a custom bookmark icon.  This way if someone saves the bookmark to thier home screen it stands out.  The two site below cover how to do this

http://eisabainyo.net/weblog/2009/06/12/making-a-website-iphone-friendly-using-css/

http://stackoverflow.com/questions/1951381/configuring-android-web-applications

In short you need to have these two tags

<link rel="apple-touch-icon" href="../../apple-touch-icon.png" />
<link rel="apple-touch-precomposed" href="../../apple-touch-icon.png" />

Inspirations

A great site for mobile app inspirations is http://cssiphone.com/.  I attempted to make my site look like these (didn’t really succeed however.)

South Florida Code Camp 2010 Presentations are up

I have posted the presentations Chris Taylor and I did at the South Florida Code Camp.  You can find them here

http://ulti-swat.wikispaces.com/SwatIntroDemos

Specifically

An Introduction to UI testing using SWAT

Applying modern software development techniques to automating the web UI

Please let me know any feedback or questions you might have.