|
|
Introduction
This page describes certain known problems, workarounds,
and other limitations of the current eValid™ version.
LAN-Based Script Files
Windows imposes limits
which vary with machine configuration on
the total number of copies of a script file
that can be open over the LAN,
i.e. not local to the machine on which eValid is executing.
This may be seen as a "sharing violation" during load test runs.
To avoid this it is good practice to run eValid
with scripts that are in files local to the current machine.
Underlying Operating System
The current releases of
eValid are fully supported on these Windows versions:
2000 (including server editions); XP; Vista; Windows 7 (W7); Windows 8 (W8).
eValid may work on Windows 98/ME/NT with some problems;
however, the eValid product is not supported on those systems.
Java Applet Version Issues
The eValid engine is optimized to run with the Microsoft JRE Runtime environment.
eValid also operates with Sun's JRE but certain commands that work correctly
with the Microsoft JRE will not work with the Sun JRE due to limitations that
have been imposed by SUN.
Link Check Wizards
The static link check wizard has capacity limits and a page with a large number of links may exceed the
capacity of the script dialog window.
The dynamic link check, which creates links "on the fly," doesn't have a capacity problem but
may attempt to create naviations to pages that occur only as PopUps or which involve lengthly
downloads.
Caution is advised.
Whitespace Processing
eValid assumes "well-formed HTML" and is known to have difficulties with HTML
compositions that include complicated internal layouts.
The most common issue arise when whitespace (tabs, newlines, CR/LF's, and blanks)
is used to format the HTML for internal readability purposes in a way that affects
the actual DOM content when the page is finally rendered.
In these cases manual correction of recorded scripts may be needed.
Parameter Processing
Certain limits apply to how parameters are passed from a parent script
(e.g. via the CallScript ... or GoScript ... commands) to a target script.
A data synthesis/substitution line consists of variable declarations and
its definitions for one playback in the script.
For example, consider the substitution:
$NAME=George $CITY=SF
Tabs and spaces are removed from each line; thus this string:
$CITY=San Francisco
will define a variable $CITY with the value
(without the quotes):
"SanFrancisco"
A dollar sign ($) in the data substitution line will always declare a new variable. An equal sign (=) after the declaration always defines the variable.
Session Cookies
Some websites use session cookies (cookies that are created by the server and
kept by the eValid browser but are not put in the cache) to manage security.
In some cases session cookies are not handled properly when the server
expects the browser to behave according to the IE 6.n level of support.
If dynamic session cookies appear to be dropped during playback when you are using
eValid with an IE 5.n base an upgrade to IE 6.n will solve this problem.
Detailed Timing
Detailed timing data is derived by eValid interacting with cache at the time a playback activity that
involves a page download.
Results may not be accurate in case more than one eValid is accessing the cache at the same time.
There is only one cache on the machine and multiple eValid's can access it, with possibly
erroneous results,
unless this is prevented by using the Lock ... Unlock command.
In some cases eValid will report a page element as Cached rather than Downloaded in the Event Log when detailed timing is turned on. eValid samples the cache at a 20 Hz rate (50 msec interval) and if a page fragement arrival finishes in the interval between the last time the sampling is done and the time when the "page download complete" signal is received by eValid, these pages are labeled as Cached
Mouse Sensitivity Issues, Disappearing Window
Sometimes a Windows Application is sensitive to "mouse down" events,
rather than "mouse down/mouse up" events.
If such a window resides over the eValid window and has focus,
and if a mouse down event to that window causes the window to disappear,
then the eValid window below it will gain focus (the mouse is still down!)
and the eValid window will respond as if there had been a mouse click.
Usually this results in an erroneous playback script.
The workaround in this very unusual case is to: (1) move the window that appears on top of the eValid display to a different area of the screen, or, mimimize the eValid window; and then (2) perform the required click on the window. This technique should prevent this kind of click-crossover error.
Nested Frames
eValid can handle nested frames perfectly well so long as all frames are
accessed from the same server.
If a frame nested below the top frame is served from a server different
than its parent eValid may not be able to record activity within the
nested frame correctly.
This is a security limitation
imposed by the underlying IE browser technology.
Desktop Focus Limitation
Windows has only one desktop, and two or more co-executing copies of eValid may attempt
to compete for focus on the desktop.
We recommend the use of the Lock and Unlock commands around the portions
of scripts that require exclusive access to the desktop and that may be running in parallel with
other scripts that also require focus on the desktop.
LoadTest Capacity
Licensing for LoadTest is on a per-machine basis: you can run as many eValids as you wish
on the licensed machine.
Your actual capacity will vary depending on playback script properties, total available RAM,
size of the video card RAM, and certain operating system parameters.
Site Analysis Timings, Page Size Calculation
During Site Analysis if you have chosen "Skip Top Page Refresh" in
the Site Analysis Preferences menu,
the page byte counts
for download of the top page may be zero.
This happens because the page size calculation done by eValid
is based in part on analyzing contents of the cache.
Without re-downloading the top page,
and assuming all of the parts of the top page
are present in the cache,
eValid will not need to download any new bytes and thus
will report [accurately] 0 bytes downloaded for that page.
This problem is not present if you choose NOT to "Skip Top Page Refresh".
Site Analysis Report Ordering
The algorithm used by the eValid Site Analysis engine adds URLs to its worklist in the
order in which it finds them within the current page; the spider crawls over the
worklist in that order.
However, parent/child dependencies that are round rarely form any pattern.
Instead of placing URLs in an apparently random order in the Complete Map or Unique URLs Map,
they are put in this order: (1) All Anchor Tags; (2) All Script References; (3) All IMG references.
Site Analysis Searching
The eValid SiteMap engine examines each URL string to determine
if it is searchable for additional links.
The following types of objects are not searched:
Some types of opaque objects are not currently scanned by the Site Analysis analyzer; examples are Flash objects and Java Applet objects.
Site Analysis Link Counting
There are different views available
after eValid Site Analysis is complete,
and in some cases, depending on the data and settings being used,
there can be discrepancies in link count and URL count
in the various displays.
This is due to the fact that
each display is a specific kind of view and
all of the displays selectively choose NOT to include
certain data from the whole network of
URL's and their interconnecting links.
For example, the 3D-SiteMap applet does not show data from URLs that are labeled as unvisited, excluded (for any reason), offsite, or with "status unknown".
Two Layer Modal Dialogs
eValid V9 is not able to record from a second layer within a modal dialog.
This occurs, for example,
if you are trying to record an ftp download sequence
that saves the dowloaded object to a local file.
The type in of the filename usually appears
in a sub-window to the parent modal dialog window
and cannot be recorded.
Hand Modified Scripts Failure Warning
Every effort is made in eValid to assure that script commands
do not have any negative consequences or side effects.
The general principle used in regard to any particular action is,
if it could cause a problem,
then issue an error message and avoid that action.
However,
this is not foolproof and
it is quite possible to create scripts
by hand and or by editing recorded scripts,
that produce very unusual effects.
Mixed Mode Recording
Problems can arise when recording "mixed mode" scripts, i.e. those that
involve a parent and one or more child windows, modal dialogs,
and desktop windows.
The problems you'll see probably are due to trying to have eValid
record in too many modes at the same time.
For example, before switching to DeskTop mode, it is always
best to switch all OTHER recording modes off.
Multiple Sub-Window Problems
Normally each mouse click results in a single browser response.
There are cases where a single mouse click results in both a
Navigation and a creation of a New Window
(either a new browser window or a new application window).
Basically if the site you are analyzing obeys the
"one click equals one New Window" rule eValid runs perfectly.
Some websites are set up so that a single mouse click causes creation of two or more New Windows. In this situation eValid may not make a completely reproducible recording: the window closure process could be defective. The workaround to record closing all but one of the windows during the recording. Doing this will assure that the windows close correctly. In some cases it may also be necessary to hand-edit the script.
Multiple Sub-Windows Format Explanation
When a parent window launches a sub-browser window,
during recording [and during playback]
eValid launches sub-browsers that do not have the regular controls.
This is done to make sure that,
during a test recording process,
you don't by mistake use the controls on the
eValid sub-browser to control recording or playback.
Note that you still have browser control of sub-browser windows
using a mouse right click.
Recording Sub-Windows Limits
The eValid browser can record activity in up to 32 parent-browser
launched sub-windows.
eValid 4 build versions after #45 (04:00:45) have this limit raised to 128.
Validating PDF Documents
Because PDF documents shown in the browser window
have their own ability to capture mouse and keyboard events,
it is not possible to use
Validate & Synchronize > Screen Window
on such pages.
Browser Capability
The eValid browser tracks IE technical capability.
See the
Browser Comparison Chart
for details.
Link Resolution
eValid V9 uses a local resource DLL to store almost all of its images.
This results in a more compact product that behaves more reliably
on machines on which eValid is actually installed.
However, in some cases eValid generated HTML reports will not
render correctly on machines on which eValid is not installed.
If this becomes a problem the solution is to install eValid on
the target machine.
The eValid engine is freely installable on any environment
and can be downloaded from the standard FTP address.
