Re: efs-hang on NT

Holger> Hi efs-folks ! Holger> I've reported a reproducible hang (for me, at least) on NT4SP4, with the Holger> current efs-package for XEmacs 21. That is, efs-version 1.18/1.57. Holger> First hint: the ftp-client is "eingedeutscht", i.e. it answer for Holger> example "Hash EIN (2048 Bytes/sec)" instead of using "hash mark". Holger> Fiddling with efs-hash-mask-msgs didn't help. You mean `efs-hash-mark-msgs'. JV> Does this patch help (It basically trying to push the NT ftp JV> client into flushing the buffer the same way as at other points). Holger> No, unfortunately it doesn't. But I can show where it hangs - in the Holger> same function, i.e. efs-raw-send-command in You mean `efs-raw-send-cmd'. Holger> (prog1 Holger> (if nowait Holger> nil Holger> ;; hang around for command to complete Holger> ;; Some clients die after the command is sent, if the server Holger> ;; times out. Don't wait on dead processes. Holger> (efs-kbd-quit-protect proc Holger> (while (and efs-process-busy Holger> ;; Need to recheck nowait, since it may get reset Holger> ;; in a cont. Holger> (null efs-process-nowait) Holger> (memq (process-status proc) '(run open))) Holger> (accept-process-output proc))) Holger> It jumps to the else part and then happily loops. Introducing a Holger> time-out to accept-process-output didn't change that. This really looks like a bug in the NT subprocess support. -- Cheers =8-} Chipsy Friede, Völkerverständigung und überhaupt blabla