Issues.00284 - lordmundi/wikidoctest GitHub Wiki

00284: remote commanding server doesn't play well with client side timeouts

« 00283 | Issues | 00285 »

Summary: remote commanding server doesn't play well with client side timeouts

Created: 2010–11–01 13:59

Status: Released

Category: Bug

From: frankie

Version: 2.1

Released_In: 2.2

Description:

currently, if the client side of a socket times out due to EDGE taking to long, the EDGE RCS will still try to return information on the socket even though it isn't valid.

This should be wrapped in a catch in case this condition occurs.


Comments

frankie November 18, 2010, at 06:14 PM: I was able to reproduce this bug before the fix by sending "after 4000; return foo" as the remote command on a socket with a timeout of 3 seconds.

frankie November 29, 2010, at 03:00 PM: Ok, the RCS now performs a catch when printing to a socket in case it is no longer valid by the time it tries to print.


« 00283 | Issues | 00285 »

Associated Commits

| commit | 61dcbbbc420c70398f6293b135d372bce84fa911 link5 | || | Author: | Frank Graffagnino | | Date: | Thu Nov 18 18:15:09 2010 -0600 | | Message: | [@Issue 00278 and 00284: Updated RCS to be tolerant of client dropout Now, if a client drops the socket before the server is done printing and closing it itself, it doesn't generate tcl errors from printing to a dead socket. Now, prints are wrapped in a catch. @] |

Affected Files:

gui/remote_commanding_server.tcl | 49 ++++++++++++++++++++++++++++++±--------
 1 file changed, 38 insertions(+), 11 deletions(-)