Well, after the exchanges regarding gkclient, and although the results are all the expected for this simple patches, I'll summarise my tests in case they are useful for anyone.
I've done and tested this things to avoid my gkerver/gkclient problems. I expected no problems, and did not find any.
1.- Roll back 1.23&24 in gkclient.cxx against h323plus v1.25, basically unconditionallizing again the only endpointIdentifier assignement. All tests ok, as this leaves v1.25 behaving more or less like v1.24
2.- Apply the 1.48 in gkclient.cxx against v1.25, making it clear the endpoint identifier on full registration required. All tests ok again. This is the one I'll stick to while waiting for v1.26, as I expect to be able to get rid of it in the future.
3.- Both 1 and 2. I knew they are redundant but wanted to assert they worked in the test setup, all good again.
Also I modified my gatekeeper as I posted to make it conditionally accept the endpoint identifier passed from the endpoint. All tests above worked with and without accepting endpoint identifiers. I'll keep this patch in case I encounter and endpoint with the unpatched v1.25 behaviour, as some of my endpoints are closed source.
Bear in mind all tests are done in a simple scenario of a single ras only gatekeeper with endpoints registered on it, rebooting endpoints and gatekeeper in the middle.
Francisco Olarte.