-- Nicolas Trown (ntrown@corp.inprise.com) Software Engineer - VisiBroker development Inprise Corporation http://www.inprise.com
-- BEGIN included message
- To: "'trown@ecst.csuchico.edu'" <trown@ecst.csuchico.edu>
- Subject: netrek: possible bug
- From: "Livingston, Ross" <Ross.Livingston@fmr.com>
- Date: Mon, 10 Aug 1998 20:18:13 -0400
- Delivered-To: trown@spam.ecst.csuchico.edu
- Delivered-To: trown@ecst.csuchico.edu
Nick, I've compiled netrek (vanilla) source at work and think I've found a bug. As far as I've been able to tell, you're the code owner. If you aren't, please let me know who is. Here's the problem: clue is defined as an extern int on lines 55 and 118 of ntserv/data.h It appears this variable is indeed used for two different things, and should have been given different names. I'm not terribly familiar with the code, but it seems clue is first set when the .sysdef file is parsed. clue is set to the value of CLUE to indicate if the server should check for cluedness (new word?) in sysdefaults.c (line 91). clue is later modified in time_access in timecheck.c but I think mods you made when you re-wrote code for the .time file make this modification obsolete. Anyway, the side effect is that setting CLUE=1 in .sysdef SHOULD do minimal clue checking (only ask twink to send word to self) but instead, clue has been incremented a bunch of times (as side effect of time_access() being called) and the more advanced question clue checking is used. Let me know what you think. Am I stoned? -Ross
-- END included message