User Forum Posts

Beginning in mid-2010 we have directed all technical support questions to the eValid User Forum. We have learned that when one user has an issue, all users can profit from the answer.

Here is an additional selection of some of the posts that we think would be of general interest.

• The Comparative Level of Effort of eValid Recording... -- Discusses the relative level of effort of using eValid versus other kinds of technical approaches.
• Comparing An Actual Value At Some DOM Element -- Explains how to compare an actual value in a page by reading and manipulating the current contents of the DOM.
• eValid Faithfully Emulates An IE Browser -- Discusses how we made sure that eValid was as close to IE behavior as possible.
• Can eValid Test An Application That Involves Two Browsers? -- Describes conditions under which two [or more] eValid instances can operate at the same time, in combinations of record AND playback.
• DOM Values -- How to use eValid commands to extract, save, and insert DOM values from one test context to another.
• Wht is your structural testing really good at? -- Describes when the structural testing methods are needed, and how effective they are.
• About "SyncOnText" -- Describes how eValid AJAX synchronization logic works, and what the overhead is likely to be when you are waiting for a DOM feature to synchronize AJAX activity.
• Query Strings -- Explains how to deal with URLs that have complex query string (client side parameters).
• A Different IP Address -- Details on when eValid can "spoof" being on a different IP address.

Please Contact Us for additional details about any of these topics.

Recent and Upcoming Conferences & Technical Community Events

Here are links to some technical meetings that may be of interest to WQN readers:

• 2nd International Workshop on the Quality of Service-Oriented Software Systems (QUASOSS'10), 4 October, 2010, Oslo, Norway
• 5th International Workshop on Systems Software Verification (SSV'10), 6-7 October 2010, Vancouver, BC, Canada.
• 8th CREST Open Workshop: Advanced Mootation* Testing 27 October 2010, University College, London, England (* "Mootation Testing" is a second order mutant of "Mutation Testing")
• 2nd International Workshop on Software Aging and Rejuvenation (Part of ISSRE 2010), 1 November 2010, San Jose, California, USA.
• 22nd IFIP International Conference on Testing Software and Systems (ICTSS), 8-10 November 2010, Natal, Brazil.
• 2010 Workshop on Verification, Validation and Uncertainty Analysis in High-Performance Computing (VVUHPC 2010), 14 November 2010, New Orleans, LA, USA.
• The Future of Software Engineering Symposium 22-23 November 2010, ETH, Zurich, Switzerland.

Community events like these deserve very careful consideration!

High BU-Count Server Performance Testing Projects

eValid users are already aware of the use of multiple eValid instances -- called Browser Users (BUs) -- to generate 100% realistic load on servers. Each BU runs as a separate Windows process, and each can handle very complex applications including AJAX and database-related applications.

To get high BU counts requires that we use two methods:

• The first method involves making OS parameter changes to increase the size of the virtual memory, increase the desktop heap size, and change the performance option to optimize for background operations.
• Even with these adjustments, each desktop user still runs up against a limit on the number of GDI Objects per user session. To overcome this constraint, we open additional user accounts on the driver machine.

Using EC2 cloud machines to drive very heavy BU-count loads, we usually employ 8-core High-Memory and/or High-CPU machine images. One such machine, properly configured, yields 1,000+ BUs per machine depending on the test duty-cycle and machine resource use. All of these methods come together in our PerformanceTest projects, which offer load generation up to 100,000 BUs in parallel,

User Forum Posts

Here are some eValid User Form posts of interest to the greater eValid community:

• Sync On The Same Document Object Property -- When two scripts synchronize on the same object property there is no interference because there are separate processes involved.
• The Value Of "sourceIndex" And "elementValue" -- How to use the values of these two critical DOM-based variables inside your current script.
• How Do You Do Step Functions In Load?" -- What's the best way to ramp up load when running multiple eValid instances, and how to arrange for a "step function" in that ramp schedule.
• Can Two eValid's Be Running At One Time? -- eValid technology allows multiple instances to run in parallel and not interfere with each other at all.
• Does eValid use HTTP Traffic in Simulating User Load? -- The way eValid generates load exploits the browser's ability to download pages, but there is an alternative utility that can drive simple non-brower HTTP traffic if you wish.

Please Contact Us for additional information about any of these topics.

The Fuss About ID Tags

One of the biggest advantages eValid has is its Adaptive Playback feature. This feature compensates for non-consequential web page changes by pivoting on page element's HTML ID Tags structure. The basic idea of an ID Tag is to insert something "constant" into the page HTML, and most web page editors and production systems assure fixed IDtags and assure that they are kept unique (although uniqueness is not a requirement). The ID tags were recorded into the original page, and so long as they don't change, Adaptive Playback makes sure the test runs OK. To assure full disclosure, however, eValid always reports successful use of Adaptive Playback as a non-fatal Warning message in the test's event log.

A simple example is an HTML anchor tag -- just a regular dynamic link -- where the URL and/or visible text can change but eValid can still work correctly even when the visible text changes or when the underlying URL changes -- so long as the ID tags are left alone.

When ID Tags Change: But, if the ID tags on a page DO change there's nothing to pivot on and the Adaptive Playback feature will either fail or at best give a misleading result. The test that relies on constant ID tags, in turn, will also fail. Not the best possible outcome at all.

In fact there are many applications where ID tags change and for good reason. There are some environments where the ID tags are automatically generated with new values every time the page is edited and re-published. In THAT case, pivoting tests on ID tags clearly won't be effective. Tests that ought to be OK now show false negative results.

Structural Testing Enters The Picture: In this case, a slightly stronger mode can be used to provide protection for tests even when ID tags are non-constant. What such tests do is rely on Structural/Algorithmic Testing -- where you use as a pivot any feature of the web page that know WILL be constant over time. While it takes a bit longer to create such a "structural test" the benefit is a playback that continues to reliably indicate a valid page even when the internals change.

Recent and Upcoming Conferences & Technical Community Events

Here are links to some technical meetings that may be of interest to WQN readers:

• Testing: Academic and Industrial Conference -- Practice and Research Techniques (TAIC PART 2010), 3-5 September 2010, Windsor, UK.
• 26th IEEE International Conference on Software Maintenance (ICSM 2010) , 12-18 September 2010, Timisoara, Romania.
• 1st International Workshop on Evolution Support for Model-Based Development & Testing , 16 September 2010, Ilmenau University of Technology, Erfurt, Germany, EU.
• 7th International Workshop on Web Services and Formal Methods , 16-17 September, Stevens Institute of Technology, Hoboken, New Jersey, USA
• 25th IEEE/ACM International Conference on Automated Software Engineering (ASE 2010), 20-24 September 2010, Antwerp, Belgium, EU.
• 8th International Symposium on Automated Technology for Verification and Analysis (ATVA 2010), 21-24 September 2010, Singapore.
• 4th International Workshop on Testing, ANalysis, and Verificatino of Web Software (TAV-WEB 2010), 21 September 2010, Antwerp, Belgium, EU.
• Model-Based Testing and Test Automation, 27 September 2010 - 2 October 2010, Leipzig, Germany.
• Workshop on Trustworthy Self-Organizing Systems, TSOS 2010, 28 September 2010, Budapest, Hungary, EU.

  Events like these deserve very careful consideration!

Current User Forum Posts

Beginning in mid-2010 we have directed all technical support questions to the eValid User Forum. We have learned that when one user has an issue, all users can profit from the answer.

Here is an additional selection of some of the posts that we think would be of general interest.

• Sync On The Same Document Object Property -- When two scripts synchronize on the same object property there is no interference because there are separate processes involved.
• The Value Of "sourceIndex" And "elementValue" -- How to use the values of these two critical DOM-based variables inside your current script.
• How Do You Do Step Functions In Load?" -- What's the best way to ramp up load when running multiple eValid instances, and how to arrange for a "step function" in that ramp schedule.
• Can Two eValid's Be Running At One Time? -- eValid technology allows multiple instances to run in parallel and not interfere with each other at all.
• Does eValid use HTTP Traffic in Simulating User Load? -- The way eValid generates load exploits the browser's ability to download pages, but there is an alternative utility that can drive simple non-brower HTTP traffic if you wish.