Hei Pekka, kiitos vastauksesta.
Bonjour Laurent, je vous remercie pour votre réponse.
As you mentioned, Pekka, the Jemmy documentation doesn't help us develop new libraries upon it and also there isn't the guarantee that it can test GEF components. So we decided to use the SWTBot and, as such, the EclipseLibrary.
We succeeded in using the eclipsebot.bat and the swtbot.bat approach.
We also succeeded in using the eclipselibrary-x.x-with-dependencies-swt.jar and the standalone version.
What we did not succeeded was to run the test in a new Eclipse installation.
In this case, the Start Eclipse keyword fails to start Eclipse.
After some banging on the wall, we discovered that we have to first open Eclipse through the eclipse.exe program (or through the command line java -jar eclipse/plugins/org.eclipse.equinox.launcher_1.0.0.v20070606.jar), because this allows the configuration directory (e.g. C:\Program Files\eclipse\Indigo\configuration) to be properly filled with the plugins data.
After this first start-up, the RF test succeeds.
Since the eclipse.exe is just a wrapper for the complete command line to start Eclipse, we tried to feed some of these command line options to the keyword Start Eclipse (e.g. -configuration file:/C:/Program Files/eclipse/Indigo/configuration), but with no luck.
We also tried using the keyword Start Application In Separate Thread and not only it has the same problem described for keyword Start Eclipse, but it also makes the remaining keywords of the test case to fail, because then the widgets can not be referenced.
Another test we made was to use the Start Application keyword.
In this case, Eclipse starts properly, even if it is the first time, but the widgets can not be referenced and so the test also fails.
We looked to the EclipseLibrary keywords source code, but it doesn't reveals a reason for this problems.
- Why starting an Eclipse instance in another thread requires it having the configuration directory filled?
- Why starting an Eclipse instance with the Start Application In Separate Thread keyword does not produces the same result as with keyword Start Eclipse ?
Laurent, can you please help us find the answers for these questions ?
Meanwhile we solved this problem by first starting Eclipse through the SWTBot command line and then we close it with a SWTBot test.
After this we can use the EclipseLibrary to run our RF tests.
Not an elegant solution, in my opinion...
We are also planning on extending your library to support GEF, since SWTBot 2.1.0 has this functionality and we need it to achieve our goal.
Kiitos / Merci,
Ricardo de Matos
忘れていない
On Fri, May 3, 2013 at 6:37 PM, Laurent Carbonnaux
<laurent.c...@gmail.com> wrote:
Hi All,
EclipseLibrary is still alive!
The fact it's seams not is that I didn't provide any new enhancement for a while, because I didn't need for my self use, and really no much time.
I at least give support to people who need help using it.
So if you want more, please ask for or help me develop it.
Regards,
Laurent.