On Saturday 04 January 2003 03:58 pm, Eric Pouech wrote:
o Failure in the child process doesn't cause failure of the test (is there someone else who has this solved for their test whose code I can borrow?)
couldn't you test the output of GetExitProcess (and ensure that when a test fails, we somehow return an error code) ?
yes. (duh :) )
o rpcss.exe.so runs from the INSTALLED wine, not from the build tree!! This is totally unacceptable, but I'm not sure what would be a wise way to solve the problem. I need to change the way rpcss gets invoked by rpcrt4.dll. The best I can think of so far, would be to provide some global override (via a function call, I guess) for where rpcss.exe.so should be found, and then use this "feature" to ensure that during testing, rpcss.exe.so is invoked from the build tree instead of from the PATH environment variable. Any thoughts on this?
what about an environment variable ? we already use something like this for the wineserver (WINESERVER), the wine console (WINECONSOLE)... why not rpcss (WINERPCSS) ?
hmmm... will take a look at those, thanks.
o There's a way to tell midl to deal with these namespace issues automagically so that the same .exe can be it's own client and server without a namespace conflict.... anyone know of such a feature?
there isn't is ATM however, MS' midl has a /prefix option that does just what you need sounds like the way to go
neato. sounds like a guaranteed winner :)
(another solution would be to put either client or server in a DLL (in fact a wine builtin DLL) so that you separate the name space between to two
nice, also, maybe nicer... but surely this does not belong in "./dlls"? Will the makebeastie let me put it somewhere discreet like "./dlls/mydll/tests/dlls"?
I kind of like the latter, also, because it allows me to kick around gazillions of conflicting namespaces, if need be.
Anyhow, between the two of these I will surely find a winner. Thanks so much! You know how it goes, the mental bugzapper gets silently shut down by the Universal Power Glitchifier sometimes, I think it's back up again...