[PATCHES] Re: (call-process ...) to return correct exit-status on NT

Hello Jonathan, Craig, Bill! Jonathan> call_process_cleanup(). You should probably change Jonathan> wait_for_termination() so that on win32 its argument is Jonathan> "phandle" rather than "pid". Done. if (os_subtype == OS_WIN95) { I negate the pid again. Craig, could you please test one of the attached patches again? For your convenience I have produced and tested two of them, one against 21.1.7 and one against latest CVS status past 21.2-b19. I hope this will work on Windows95/98 finally. Jonathan> How about using WaitForSingleObject(pid, 1/10s) instead Jonathan> of GetExitCodeProcess() in your do...while loop in Jonathan> wait_for_termination(). That might make XEmacs to use a I'm doing the following, which should be equivalent: QUIT; /* * Wait for the child process to be signalled. We wait * indefinitely, since spinning a loop while (ret == WAIT_TIMEOUT) * would amount to the same. */ ret = WaitForSingleObject(pid, INFINITE); Should it be possible to keyboard-quit (C-g) a long-running process? I cannot interrupt (apply 'call-process "sleep" nil nil nil (list "30")) short of using Task Manager to "End Process". Jonathan> lot less CPU and let its child process exit more Jonathan> quickly. You'll obviously still need to call Jonathan> GetExitCodeProcess() once after the child has really Jonathan> terminated. Done that. Jonathan> On Windows 95/98 process IDs are negative(!) which Jonathan> screws up in a call to make_int() in Jonathan> Fcall_process_internal() by blowing the 30(?) bit lisp Jonathan> integer size. There's a nasty hack in create_child at Jonathan> line 450 of ntproc.c to get round this by making the PID Jonathan> positive. The problem that Craig was experiencing with Jonathan> your original patch may have been due to the fact that Jonathan> you weren't renegativising the PID in the call to Jonathan> OpenProcess(). Your code still has this problem. You Attached patch is trying to fix this. Let's see what Craig (or others) will report. I am testing this on WindowsNT Workstation 4sp4 with VC++6.0. Jonathan> either have to include ntheap.h and make the pid passed Jonathan> to OpenProcess negative if (os_subtype==OS_WIN95), or Jonathan> you have to come up with a better general solution to Jonathan> the negative pid problem. I tried a drastic change to pass around process handles instead of PIDs but failed miserably. The PID concept is pretty well wired into XEmacs from the UNIX world. Jonathan> Jonathan. Thanks for all the help! I have attached Please test any of the following patches and give me feedback. TIA, Adrian Jonathan> -- Jonathan> Jonathan Harris | jhar(a)tardis.ed.ac.uk Jonathan> London, England | Jonathan.Harris(a)symbian.com