Home » Archived » Test and Performance Tools Platform (TPTP) » Automated GUI Recorder issues
| Automated GUI Recorder issues [message #52558] |
Thu, 02 February 2006 09:35  |
 Shlomy Reinstein
Messages: 9
Registered: July 2009 |
Junior Member |
|
|
Hi,
I'm using the AGR version that comes with the stable build of TPTP 4.2.0
(200512011334). I have encountered several issues when working with the AGR,
please let me know if I should open bug reports Bugzilla:
1. It fails to resolve the preference tree in the Window -> Preferences
dialog. I noticed that there is no XML entry for "Tree" in the default
adaptive widget resolver configuration.
2. If I use position-based recording to work around the previous problem, it
records "mouse-down" and "mouse-up" events when I click a preference in the
preference tree. During playback, I see the preference being selected (i.e.
has the background of a selected item), however the preference page is not
changed to that preference, and the runner keeps waiting until finally
timeout occurs. If I manually move the mouse during this wait period, the
preference page appears and the runner continues normally.
3. Selections from the pull-down menu of the "New" toolbar button (the one
that's normally under "File") cause a CoreException during playback. Here's
what it recorded for a selection of the "new class wizard":
<command type="select" contextId="toolbar"
widgetId="contribid/newWizardDropDown" x-coord="0" y-coord="22" detail="4">
<selected-item path="org.eclipse.ui.actions.NewWizardMenu"/>
</command>
The exception is thrown for the "select" line.
4. "Quick run" sometimes fails to find the verification hook class, and
throws an exception that shows only the "startup.jar" on the class path.
When debugging it (I set the DEBUG flag in the execution object to true in
order to connect with a debugger), I see that the class path is created fine
(except a small problem with the first entry), but for some reason the class
loader still fails to find the class.
5. At other times, "Quick run" uses a verification class from the wrong
project. That is, I have several plug-in projects with AGR test suites, and
each has a class named "VerificationClass" that is used by the test suites
in that plug-in. When running a test case from one of these projects, the
verification class used by "Quick run" might be taken from another project
(and then, of course, either the method is not found or the wrong method is
invoked...).
6. Sometimes, the AGR runner fails to find the view that should be provided
to a verification hook although the test plug-in has view's plug-in as a
dependency and I can see the view during playback. This usually happens when
attach a debugger (to debug the verification hook).
I also have a question regarding the AGR: Is it possible to run several AGR
test suites sequentially (or a combination of AGR and JUnit test suites)
using a single launch configuration, e.g. by defining a deployment? If not,
is there a plan to enable this?
Thanks,
Shlomy
|
|
|
| Re: Automated GUI Recorder issues [message #52639 is a reply to message #52558] |
Thu, 02 February 2006 17:47  |
No real name
Messages: 404
Registered: July 2009 |
Senior Member |
|
|
Hi Shlomy,
Please see my comments below:
Hope this helps.
"Shlomy Reinstein" <sreinstein@e-sim.co.il> wrote in message
news:drsjrb$fqv$1@utils.eclipse.org...
> Hi,
>
> I'm using the AGR version that comes with the stable build of TPTP 4.2.0
> (200512011334). I have encountered several issues when working with the
AGR,
> please let me know if I should open bug reports Bugzilla:
> 1. It fails to resolve the preference tree in the Window -> Preferences
> dialog. I noticed that there is no XML entry for "Tree" in the default
> adaptive widget resolver configuration.
I tried this in quick mode with a newer build TPTP-4.2.0-200602020100 and
didn't experience any problem. (Note: you won't be able to run
test suite in standard mode with this build because of bug #:
https://bugs.eclipse.org/bugs/show_bug.cgi?id=126219)
Did you generate the macro with the build that you tried or was it
originally generated with an older build?
If the macro was generated with an older build, then please attach the full
macro to this thread. I did make some
changes to better resolve menu items, but I was careful to support backward
compatibility. If feasible, can
you also try this with a newer build?
> 2. If I use position-based recording to work around the previous problem,
it
> records "mouse-down" and "mouse-up" events when I click a preference in
the
> preference tree. During playback, I see the preference being selected
(i.e.
> has the background of a selected item), however the preference page is not
> changed to that preference, and the runner keeps waiting until finally
> timeout occurs. If I manually move the mouse during this wait period, the
> preference page appears and the runner continues normally.
There were issues with position based recording that I have already
resolved. You will
not experience the same problem if you use a newer build.
> 3. Selections from the pull-down menu of the "New" toolbar button (the one
> that's normally under "File") cause a CoreException during playback.
Here's
> what it recorded for a selection of the "new class wizard":
> <command type="select" contextId="toolbar"
> widgetId="contribid/newWizardDropDown" x-coord="0" y-coord="22"
detail="4">
> <selected-item path="org.eclipse.ui.actions.NewWizardMenu"/>
> </command>
> The exception is thrown for the "select" line.
Again, I didn't experience any problem with the build that I mentioned in
the beginning.
> 4. "Quick run" sometimes fails to find the verification hook class, and
> throws an exception that shows only the "startup.jar" on the class path.
> When debugging it (I set the DEBUG flag in the execution object to true in
> order to connect with a debugger), I see that the class path is created
fine
> (except a small problem with the first entry), but for some reason the
class
> loader still fails to find the class.
This is an odd problem but I think 4) and 5) may be related. Look at the
generated macro
and find the verification command that corresponds to your verification
hook. An example
of a fragment that corresponds to a hook is:
<macro version="0.1" >
<shell id="org.eclipse.ui.internal.WorkbenchWindow" return-code="-1">
<command type="verification"
contextId="view/org.eclipse.hyades.trace.internal.ui.PDProjectExplorer "
location="/Plugin/src" resource="plugin.AutoGUITest"
hook="verifyState:Qorg.eclipse.ui.IViewPart;"/>
<command type="wait"/>
</shell>
</macro>
Ensure that the location and resource attributes are set properly. Make
sure that the class is
the right one that you want to execute.
If this problem is still reproducible and you can provide me with a test
suite to reproduce it, then
open a defect.
> 5. At other times, "Quick run" uses a verification class from the wrong
> project. That is, I have several plug-in projects with AGR test suites,
and
> each has a class named "VerificationClass" that is used by the test suites
> in that plug-in. When running a test case from one of these projects, the
> verification class used by "Quick run" might be taken from another project
> (and then, of course, either the method is not found or the wrong method
is
> invoked...).
See advice for (4).
> 6. Sometimes, the AGR runner fails to find the view that should be
provided
> to a verification hook although the test plug-in has view's plug-in as a
> dependency and I can see the view during playback. This usually happens
when
> attach a debugger (to debug the verification hook).
This happens when the Eclipse workbench that is running the test suite is
not given focus. Make sure that focus
is given to the Eclipse workbench running your test suite. This probably
happens more often during debug mode because the debugger
changes your focus when it hits a break point.
>
> I also have a question regarding the AGR: Is it possible to run several
AGR
> test suites sequentially (or a combination of AGR and JUnit test suites)
> using a single launch configuration, e.g. by defining a deployment? If
not,
> is there a plan to enable this?
You can use the automatable service framework that uses a light client (e.g.
ant scripts, shell scripts, and etc...) to run
tests. See Help Contents > TPTP Tester Guide > Performance Testing with
TPTP >
Launching tests from scripts and applications.
>
> Thanks,
> Shlomy
>
>
>
>
|
|
|
| Re: Automated GUI Recorder issues [message #52718 is a reply to message #52639] |
Thu, 02 February 2006 23:11 |
Shlomy Reinstein
Messages: 9
Registered: July 2009 |
Junior Member |
|
|
Hi,
Thanks for the quick reply. To reply your question for (1), I recorded and
ran it with the same build (12011334). Shortly after sending the original
email, I recorded again (using the same build) and this time it did resolve
the tree...
Shlomy
"Ali Mehregani" <amehrega@ca.ibm.com> wrote in message
news:drtgji$7na$1@utils.eclipse.org...
> Hi Shlomy,
>
> Please see my comments below:
>
> Hope this helps.
>
> "Shlomy Reinstein" <sreinstein@e-sim.co.il> wrote in message
> news:drsjrb$fqv$1@utils.eclipse.org...
>> Hi,
>>
>> I'm using the AGR version that comes with the stable build of TPTP 4.2.0
>> (200512011334). I have encountered several issues when working with the
> AGR,
>> please let me know if I should open bug reports Bugzilla:
>> 1. It fails to resolve the preference tree in the Window -> Preferences
>> dialog. I noticed that there is no XML entry for "Tree" in the default
>> adaptive widget resolver configuration.
>
> I tried this in quick mode with a newer build TPTP-4.2.0-200602020100 and
> didn't experience any problem. (Note: you won't be able to run
> test suite in standard mode with this build because of bug #:
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=126219)
> Did you generate the macro with the build that you tried or was it
> originally generated with an older build?
> If the macro was generated with an older build, then please attach the
> full
> macro to this thread. I did make some
> changes to better resolve menu items, but I was careful to support
> backward
> compatibility. If feasible, can
> you also try this with a newer build?
>
>
>> 2. If I use position-based recording to work around the previous problem,
> it
>> records "mouse-down" and "mouse-up" events when I click a preference in
> the
>> preference tree. During playback, I see the preference being selected
> (i.e.
>> has the background of a selected item), however the preference page is
>> not
>> changed to that preference, and the runner keeps waiting until finally
>> timeout occurs. If I manually move the mouse during this wait period, the
>> preference page appears and the runner continues normally.
>
>
> There were issues with position based recording that I have already
> resolved. You will
> not experience the same problem if you use a newer build.
>
>> 3. Selections from the pull-down menu of the "New" toolbar button (the
>> one
>> that's normally under "File") cause a CoreException during playback.
> Here's
>> what it recorded for a selection of the "new class wizard":
>> <command type="select" contextId="toolbar"
>> widgetId="contribid/newWizardDropDown" x-coord="0" y-coord="22"
> detail="4">
>> <selected-item path="org.eclipse.ui.actions.NewWizardMenu"/>
>> </command>
>> The exception is thrown for the "select" line.
>
> Again, I didn't experience any problem with the build that I mentioned in
> the beginning.
>
>> 4. "Quick run" sometimes fails to find the verification hook class, and
>> throws an exception that shows only the "startup.jar" on the class path.
>> When debugging it (I set the DEBUG flag in the execution object to true
>> in
>> order to connect with a debugger), I see that the class path is created
> fine
>> (except a small problem with the first entry), but for some reason the
> class
>> loader still fails to find the class.
>
> This is an odd problem but I think 4) and 5) may be related. Look at the
> generated macro
> and find the verification command that corresponds to your verification
> hook. An example
> of a fragment that corresponds to a hook is:
>
> <macro version="0.1" >
> <shell id="org.eclipse.ui.internal.WorkbenchWindow" return-code="-1">
> <command type="verification"
> contextId="view/org.eclipse.hyades.trace.internal.ui.PDProjectExplorer "
> location="/Plugin/src" resource="plugin.AutoGUITest"
> hook="verifyState:Qorg.eclipse.ui.IViewPart;"/>
> <command type="wait"/>
> </shell>
> </macro>
>
>
> Ensure that the location and resource attributes are set properly. Make
> sure that the class is
> the right one that you want to execute.
>
> If this problem is still reproducible and you can provide me with a test
> suite to reproduce it, then
> open a defect.
>
>> 5. At other times, "Quick run" uses a verification class from the wrong
>> project. That is, I have several plug-in projects with AGR test suites,
> and
>> each has a class named "VerificationClass" that is used by the test
>> suites
>> in that plug-in. When running a test case from one of these projects, the
>> verification class used by "Quick run" might be taken from another
>> project
>> (and then, of course, either the method is not found or the wrong method
> is
>> invoked...).
>
> See advice for (4).
>
>> 6. Sometimes, the AGR runner fails to find the view that should be
> provided
>> to a verification hook although the test plug-in has view's plug-in as a
>> dependency and I can see the view during playback. This usually happens
> when
>> attach a debugger (to debug the verification hook).
>
> This happens when the Eclipse workbench that is running the test suite is
> not given focus. Make sure that focus
> is given to the Eclipse workbench running your test suite. This probably
> happens more often during debug mode because the debugger
> changes your focus when it hits a break point.
>
>>
>> I also have a question regarding the AGR: Is it possible to run several
> AGR
>> test suites sequentially (or a combination of AGR and JUnit test suites)
>> using a single launch configuration, e.g. by defining a deployment? If
> not,
>> is there a plan to enable this?
>
> You can use the automatable service framework that uses a light client
> (e.g.
> ant scripts, shell scripts, and etc...) to run
> tests. See Help Contents > TPTP Tester Guide > Performance Testing with
> TPTP >
> Launching tests from scripts and applications.
>
>
>>
>> Thanks,
>> Shlomy
>>
>>
>>
>>
>
>
|
|
|
| Re: Automated GUI Recorder issues [message #54049 is a reply to message #52639] |
Mon, 20 February 2006 11:12 |
Shlomy Reinstein
Messages: 9
Registered: July 2009 |
Junior Member |
|
|
Hi,
Regarding the bug you mentioned
(https://bugs.eclipse.org/bugs/show_bug.cgi?id=126219), that doesn't let me
run test suites in standard mode: The problem is even worse than the
inconclusive verdict - the test won't launch. For me, the problem happens on
Windows 2000 SP4, not on Windows XP (i.e. on Windows XP it runs fine, and
the execution history file is created with the correct verdict).
If I to launch the test without the agent controller service, I get a
CoreException telling me to check that the agent controller is running and
is configured correctly. If I launch the test (even a simple JUnit test)
with the agent controller, the launcher just "hangs" - the progress bar
reaches 54% and hangs there. The test never begins.
The same holds both for two builds that I've checked: 200602020100 and
200602160100 (both belong to TPTP 4.2). Note that in the earlier build
(200602020100), the launcher rarely succeeds in launching the test but I get
the inconclusive verdict. Most times the test cannot even launch.
Regarding the inconclusive verdict, I've added a reply to the above bug,
indicating that it happens on Windows 2000 SP4. Should I create a new bug
for the other problem (not being able to launch tests)?
Thanks,
Shlomy
"Ali Mehregani" <amehrega@ca.ibm.com> wrote in message
news:drtgji$7na$1@utils.eclipse.org...
> Hi Shlomy,
>
> Please see my comments below:
>
> Hope this helps.
>
> "Shlomy Reinstein" <sreinstein@e-sim.co.il> wrote in message
> news:drsjrb$fqv$1@utils.eclipse.org...
>> Hi,
>>
>> I'm using the AGR version that comes with the stable build of TPTP 4.2.0
>> (200512011334). I have encountered several issues when working with the
> AGR,
>> please let me know if I should open bug reports Bugzilla:
>> 1. It fails to resolve the preference tree in the Window -> Preferences
>> dialog. I noticed that there is no XML entry for "Tree" in the default
>> adaptive widget resolver configuration.
>
> I tried this in quick mode with a newer build TPTP-4.2.0-200602020100 and
> didn't experience any problem. (Note: you won't be able to run
> test suite in standard mode with this build because of bug #:
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=126219)
> Did you generate the macro with the build that you tried or was it
> originally generated with an older build?
> If the macro was generated with an older build, then please attach the
> full
> macro to this thread. I did make some
> changes to better resolve menu items, but I was careful to support
> backward
> compatibility. If feasible, can
> you also try this with a newer build?
>
>
>> 2. If I use position-based recording to work around the previous problem,
> it
>> records "mouse-down" and "mouse-up" events when I click a preference in
> the
>> preference tree. During playback, I see the preference being selected
> (i.e.
>> has the background of a selected item), however the preference page is
>> not
>> changed to that preference, and the runner keeps waiting until finally
>> timeout occurs. If I manually move the mouse during this wait period, the
>> preference page appears and the runner continues normally.
>
>
> There were issues with position based recording that I have already
> resolved. You will
> not experience the same problem if you use a newer build.
>
>> 3. Selections from the pull-down menu of the "New" toolbar button (the
>> one
>> that's normally under "File") cause a CoreException during playback.
> Here's
>> what it recorded for a selection of the "new class wizard":
>> <command type="select" contextId="toolbar"
>> widgetId="contribid/newWizardDropDown" x-coord="0" y-coord="22"
> detail="4">
>> <selected-item path="org.eclipse.ui.actions.NewWizardMenu"/>
>> </command>
>> The exception is thrown for the "select" line.
>
> Again, I didn't experience any problem with the build that I mentioned in
> the beginning.
>
>> 4. "Quick run" sometimes fails to find the verification hook class, and
>> throws an exception that shows only the "startup.jar" on the class path.
>> When debugging it (I set the DEBUG flag in the execution object to true
>> in
>> order to connect with a debugger), I see that the class path is created
> fine
>> (except a small problem with the first entry), but for some reason the
> class
>> loader still fails to find the class.
>
> This is an odd problem but I think 4) and 5) may be related. Look at the
> generated macro
> and find the verification command that corresponds to your verification
> hook. An example
> of a fragment that corresponds to a hook is:
>
> <macro version="0.1" >
> <shell id="org.eclipse.ui.internal.WorkbenchWindow" return-code="-1">
> <command type="verification"
> contextId="view/org.eclipse.hyades.trace.internal.ui.PDProjectExplorer "
> location="/Plugin/src" resource="plugin.AutoGUITest"
> hook="verifyState:Qorg.eclipse.ui.IViewPart;"/>
> <command type="wait"/>
> </shell>
> </macro>
>
>
> Ensure that the location and resource attributes are set properly. Make
> sure that the class is
> the right one that you want to execute.
>
> If this problem is still reproducible and you can provide me with a test
> suite to reproduce it, then
> open a defect.
>
>> 5. At other times, "Quick run" uses a verification class from the wrong
>> project. That is, I have several plug-in projects with AGR test suites,
> and
>> each has a class named "VerificationClass" that is used by the test
>> suites
>> in that plug-in. When running a test case from one of these projects, the
>> verification class used by "Quick run" might be taken from another
>> project
>> (and then, of course, either the method is not found or the wrong method
> is
>> invoked...).
>
> See advice for (4).
>
>> 6. Sometimes, the AGR runner fails to find the view that should be
> provided
>> to a verification hook although the test plug-in has view's plug-in as a
>> dependency and I can see the view during playback. This usually happens
> when
>> attach a debugger (to debug the verification hook).
>
> This happens when the Eclipse workbench that is running the test suite is
> not given focus. Make sure that focus
> is given to the Eclipse workbench running your test suite. This probably
> happens more often during debug mode because the debugger
> changes your focus when it hits a break point.
>
>>
>> I also have a question regarding the AGR: Is it possible to run several
> AGR
>> test suites sequentially (or a combination of AGR and JUnit test suites)
>> using a single launch configuration, e.g. by defining a deployment? If
> not,
>> is there a plan to enable this?
>
> You can use the automatable service framework that uses a light client
> (e.g.
> ant scripts, shell scripts, and etc...) to run
> tests. See Help Contents > TPTP Tester Guide > Performance Testing with
> TPTP >
> Launching tests from scripts and applications.
>
>
>>
>> Thanks,
>> Shlomy
>>
>>
>>
>>
>
>
|
|
|
| Re: Automated GUI Recorder issues [message #54153 is a reply to message #54049] |
Mon, 20 February 2006 21:52 |
No real name
Messages: 404
Registered: July 2009 |
Senior Member |
|
|
Hi Shlomy,
I've re-opened the defect and added your comments below.
"Shlomy Reinstein" <sreinstein@e-sim.co.il> wrote in message
news:dtc88d$j7d$1@utils.eclipse.org...
> Hi,
>
> Regarding the bug you mentioned
> (https://bugs.eclipse.org/bugs/show_bug.cgi?id=126219), that doesn't let
me
> run test suites in standard mode: The problem is even worse than the
> inconclusive verdict - the test won't launch. For me, the problem happens
on
> Windows 2000 SP4, not on Windows XP (i.e. on Windows XP it runs fine, and
> the execution history file is created with the correct verdict).
>
> If I to launch the test without the agent controller service, I get a
> CoreException telling me to check that the agent controller is running and
> is configured correctly. If I launch the test (even a simple JUnit test)
> with the agent controller, the launcher just "hangs" - the progress bar
> reaches 54% and hangs there. The test never begins.
>
> The same holds both for two builds that I've checked: 200602020100 and
> 200602160100 (both belong to TPTP 4.2). Note that in the earlier build
> (200602020100), the launcher rarely succeeds in launching the test but I
get
> the inconclusive verdict. Most times the test cannot even launch.
>
> Regarding the inconclusive verdict, I've added a reply to the above bug,
> indicating that it happens on Windows 2000 SP4. Should I create a new bug
> for the other problem (not being able to launch tests)?
>
> Thanks,
> Shlomy
>
> "Ali Mehregani" <amehrega@ca.ibm.com> wrote in message
> news:drtgji$7na$1@utils.eclipse.org...
> > Hi Shlomy,
> >
> > Please see my comments below:
> >
> > Hope this helps.
> >
> > "Shlomy Reinstein" <sreinstein@e-sim.co.il> wrote in message
> > news:drsjrb$fqv$1@utils.eclipse.org...
> >> Hi,
> >>
> >> I'm using the AGR version that comes with the stable build of TPTP
4.2.0
> >> (200512011334). I have encountered several issues when working with the
> > AGR,
> >> please let me know if I should open bug reports Bugzilla:
> >> 1. It fails to resolve the preference tree in the Window -> Preferences
> >> dialog. I noticed that there is no XML entry for "Tree" in the default
> >> adaptive widget resolver configuration.
> >
> > I tried this in quick mode with a newer build TPTP-4.2.0-200602020100
and
> > didn't experience any problem. (Note: you won't be able to run
> > test suite in standard mode with this build because of bug #:
> > https://bugs.eclipse.org/bugs/show_bug.cgi?id=126219)
> > Did you generate the macro with the build that you tried or was it
> > originally generated with an older build?
> > If the macro was generated with an older build, then please attach the
> > full
> > macro to this thread. I did make some
> > changes to better resolve menu items, but I was careful to support
> > backward
> > compatibility. If feasible, can
> > you also try this with a newer build?
> >
> >
> >> 2. If I use position-based recording to work around the previous
problem,
> > it
> >> records "mouse-down" and "mouse-up" events when I click a preference in
> > the
> >> preference tree. During playback, I see the preference being selected
> > (i.e.
> >> has the background of a selected item), however the preference page is
> >> not
> >> changed to that preference, and the runner keeps waiting until finally
> >> timeout occurs. If I manually move the mouse during this wait period,
the
> >> preference page appears and the runner continues normally.
> >
> >
> > There were issues with position based recording that I have already
> > resolved. You will
> > not experience the same problem if you use a newer build.
> >
> >> 3. Selections from the pull-down menu of the "New" toolbar button (the
> >> one
> >> that's normally under "File") cause a CoreException during playback.
> > Here's
> >> what it recorded for a selection of the "new class wizard":
> >> <command type="select" contextId="toolbar"
> >> widgetId="contribid/newWizardDropDown" x-coord="0" y-coord="22"
> > detail="4">
> >> <selected-item path="org.eclipse.ui.actions.NewWizardMenu"/>
> >> </command>
> >> The exception is thrown for the "select" line.
> >
> > Again, I didn't experience any problem with the build that I mentioned
in
> > the beginning.
> >
> >> 4. "Quick run" sometimes fails to find the verification hook class, and
> >> throws an exception that shows only the "startup.jar" on the class
path.
> >> When debugging it (I set the DEBUG flag in the execution object to true
> >> in
> >> order to connect with a debugger), I see that the class path is created
> > fine
> >> (except a small problem with the first entry), but for some reason the
> > class
> >> loader still fails to find the class.
> >
> > This is an odd problem but I think 4) and 5) may be related. Look at
the
> > generated macro
> > and find the verification command that corresponds to your verification
> > hook. An example
> > of a fragment that corresponds to a hook is:
> >
> > <macro version="0.1" >
> > <shell id="org.eclipse.ui.internal.WorkbenchWindow" return-code="-1">
> > <command type="verification"
> > contextId="view/org.eclipse.hyades.trace.internal.ui.PDProjectExplorer "
> > location="/Plugin/src" resource="plugin.AutoGUITest"
> > hook="verifyState:Qorg.eclipse.ui.IViewPart;"/>
> > <command type="wait"/>
> > </shell>
> > </macro>
> >
> >
> > Ensure that the location and resource attributes are set properly. Make
> > sure that the class is
> > the right one that you want to execute.
> >
> > If this problem is still reproducible and you can provide me with a test
> > suite to reproduce it, then
> > open a defect.
> >
> >> 5. At other times, "Quick run" uses a verification class from the wrong
> >> project. That is, I have several plug-in projects with AGR test suites,
> > and
> >> each has a class named "VerificationClass" that is used by the test
> >> suites
> >> in that plug-in. When running a test case from one of these projects,
the
> >> verification class used by "Quick run" might be taken from another
> >> project
> >> (and then, of course, either the method is not found or the wrong
method
> > is
> >> invoked...).
> >
> > See advice for (4).
> >
> >> 6. Sometimes, the AGR runner fails to find the view that should be
> > provided
> >> to a verification hook although the test plug-in has view's plug-in as
a
> >> dependency and I can see the view during playback. This usually happens
> > when
> >> attach a debugger (to debug the verification hook).
> >
> > This happens when the Eclipse workbench that is running the test suite
is
> > not given focus. Make sure that focus
> > is given to the Eclipse workbench running your test suite. This
probably
> > happens more often during debug mode because the debugger
> > changes your focus when it hits a break point.
> >
> >>
> >> I also have a question regarding the AGR: Is it possible to run several
> > AGR
> >> test suites sequentially (or a combination of AGR and JUnit test
suites)
> >> using a single launch configuration, e.g. by defining a deployment? If
> > not,
> >> is there a plan to enable this?
> >
> > You can use the automatable service framework that uses a light client
> > (e.g.
> > ant scripts, shell scripts, and etc...) to run
> > tests. See Help Contents > TPTP Tester Guide > Performance Testing with
> > TPTP >
> > Launching tests from scripts and applications.
> >
> >
> >>
> >> Thanks,
> >> Shlomy
> >>
> >>
> >>
> >>
> >
> >
>
>
|
|
| |
Goto Forum:
Current Time: Fri Apr 19 21:32:07 GMT 2024
Powered by FUDForum. Page generated in 0.03323 seconds |
] 
Search
Help
Register
Login
Home
