Community
Participate
Working Groups
When exporting a product without the delta-pack, the eclipse.exe is taken from the running eclipse. However, if the running eclipse is a branded product the launcher is not named eclipse.exe and does not get included in the product. Export should try and included the branded exe
We should be able to find the name of the executable using the eclipse.launcher property as specified in bug 205832.
oooo! we should get this in for 3.4
Created attachment 88399 [details] mylyn/context/zip Mylyn context
Created attachment 88400 [details] The launcher path's is retrieved in eclipse.launch system property It should work, even if I haven't fully tested it!
Thanks Ben, we'll look at it for 3.4M5
The property "eclipse.launcher" is set by Main when the command line argument "-launcher" is specified. This argument is specified automatically by the eclipse launcher. It also seems to be specified automatically by PDE when debugging. However, if Eclipse was started using java.exe without the -launcher argument then the eclipse.launcher property will not be set. PDE should fall back and look for eclipse.exe in this case. It looks like the current patch may throw an NPE in this case.
Oops. What would be the correct behaviour to your mind ? If eclipse.launcher is here, let's use it; else, keep the current behaviour ? (In reply to comment #6) > The property "eclipse.launcher" is set by Main when the command line argument > "-launcher" is specified. This argument is specified automatically by the > eclipse launcher. It also seems to be specified automatically by PDE when > debugging. > > However, if Eclipse was started using java.exe without the -launcher argument > then the eclipse.launcher property will not be set. PDE should fall back and > look for eclipse.exe in this case. It looks like the current patch may throw > an NPE in this case. >
That seems reasonable, new patch ;p?
OK... ;) Let's go! Benjamin, a.k.a. the trivial patcher :) (In reply to comment #8) > That seems reasonable, new patch ;p? >
(In reply to comment #9) > Benjamin, a.k.a. the trivial patcher :) Anything to help things not fall through the cracks is greatly appreciated. We can get you on the PDE Contributers page :P (http://www.eclipse.org/pde/pde-ui/committers/committers.php)
Created attachment 88431 [details] Updated patch... ...with a bit of code cleaning :)
Created attachment 88787 [details] org.eclipse.pde.ui.patch An updated patch. I found an issue when you're self-hosting... we don't properly pass the -launcher when we are on win32 causing an issue when you're actually going to test this stuff in self-hosting mode ;)
Created attachment 88788 [details] mylyn/context/zip
The patch looks good to me :) Chris and I were talking and the only thing that might make this a little more robust is to check to make sure the File specified in the system properties doesn't only exist, but is also not a directory (which is why Chris had to add the launcher modification).
yap!
Created attachment 88814 [details] org.eclipse.pde.patch I updated the patch to be a bit more resilient if we hit a corner case of getting a bad launcher path. Thanks to everyone for their help on this issue.
Created attachment 88815 [details] mylyn/context/zip
done!
great :)