[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Info-xnee] Fwd: How to use Xnee to perform the auto GUI test for gt
Re: [Info-xnee] Fwd: How to use Xnee to perform the auto GUI test for gtk+ based phone applications?
Tue, 01 Jan 2008 19:27:57 +0100
Thunderbird 220.127.116.11 (X11/20071022)
li zhou skrev:
> Hello all:
> We are making contributions to LiPS ( http://www.lipsforum.org ) and
> developing some software and applications to implement the LiPS
> Specifications. Some software have been done. To ensure the quality of those
> software and save resources, we want to use some open source tools to
> perform the auto GUI test for some gtk+ based phone applications like
> address book, voice call. Xnee is a good candidate. And yes, it is indeed.
> But after some investigation, we find that Xnee can't help to determine the
> correctness of the test result except the recording and replaying. So we
> have no choice but turn to your help whether we can find a solution to
> achieve this.
> To explain our requirements clearly, here is a example of the "contact
> search" for address book application . For example, in the search interface
> of Address book, when typing 'j' through phone keypad , the matched
> contacts starting with 'j' will be listed in alphabet order. we name the
> matched contacts as the expected results.
> Our current status is: test specification is ready and we can perform the
> recording and replaying correctly with Xnee. The scenario we image is like
> (1). tester or developer perform the test manually according to the test
> specification to obtain the test script which just record the EVENT
> sequence, that is the '*recording*' process.
> (2). For the regression test, the test will be perform the test
> automatically without tester's intervention.
> The current Xnee release can't meet the requirement of step (2). After
> reading the Xnee manual, maybe the 'synchronism mechanism in Xnee" is
> helpful. As the synchronism mechanism, it records more DATA than just the
> EVENT to be replayed. Based on this mechanism, we think that can we also
> introduce this mechanism to step (1) to record the test results as the DATA
> field in the script file. (for each test case in our test
> specifications, the results/outputs will be specified for the given inputs.
> this is the base of our idea.) During the replaying process, just
> comparing the DATA field with those DATA from recording process Xnee will
> determine the correctness automatically. Because the DATA output
> will be same for each test case ideally.
> You are the exact expert,
guess not ;)
> would you be so kind to help us out on
> the feasibility and implementation. If necessary, we can make a conf-call
> with you.
Right now I see two ways to solve your problem:
1) Do small printouts in your tested program that can be enabled in
compile time through C macros (#define) or run time through a command
line option. Compare the printouts when you recorded with the printout
from when replayed.
2) Use the -ep and -ek options. I just discovererd a newliy introduced
bug in them so hold on for a short while, will fix it asap. An example:
-ek F5 -ep "do-snapshot.sh"
2.1 Make a script (do-snapshot.sh) that takes a snapshot of your
display and stores it somewhere.
2.2 Every time you press F5 during record, Xnee will
* add a mark in the recorded file (see below) together with
* launch do-snapshot.sh
2.3 Move the image somewhere else.
2.4 When replaying Xnee will eventually read the special mark in the
recorded file. When reading this Xnee will launch do-snapshot.sh.
2.5 Compare the image from the recorded session with the newly
... could any of the procedures above help you???
 one can argue and say that the program you're testing is not the
same as you 'deliver' to your customer. Make sure the printout does NOT
interfere with the programs normal work!!!
> Many thanks and looking forward to your help....
> Best Regards,
> Info-xnee mailing list
- Re: [Info-xnee] Fwd: How to use Xnee to perform the auto GUI test for gtk+ based phone applications?,
Henrik Sandklef <=