[comp.windows.news] A race condition?

fj@iesd.dk (Frank Jensen) (05/06/89)

The NeWS manuals suggest the following scheme in order to synchronize
a client with the NeWS server:

	SOME_TAG tagprint
	uniquecid typedprint
	datum_1 typedprint
	...
	datum_n typedprint
	[currentcid cidinterest] forkeventmgr
	waitprocess pop

But suppose that the client responds very fast (perhaps because it has
nothing to do), isn't it then possible that the process will be forked
too late (with the result that something will hang)?

I have experienced something that seems to confirm this (although I
must confess that I had a `setstandardcursor' command before the fork
command).

Is it said anywhere in the NeWS manuals that this can't happen?
Or is it implicitly guaranteed to be impossible?


Frank Jensen,   fj@iesd.dk
Department of Mathematics and Computer Science
Aalborg University
DENMARK