jld#@andrew.cmu.edu.UUCP (04/23/87)
The fix to update.c that corrects the "rolling civilians" bug brings
out another bug in update.c.
It should only do an update of a sector if the last time that sector
was updated is < the current turn number, not <=. The <= causes the program
go into infinite recursion if there is a loop in delivery routes (even if
each path of the loop is a different item).
Below is a diff of the update.c file which shows both bug fixes (both
are one line changes).
*** update.old Thu Apr 23 17:11:59 1987
--- update.c Thu Apr 23 17:49:37 1987
***************
*** 31,37
verbose = selup & UP_VERBOSE;
quiet = selup & UP_QUIET;
if( sect.sct_lstup == 0 ) return(0);
! if( sect.sct_lstup <= curup ) goto X160;
sect.sct_lstup = curup;
return(1);
X160:
--- 31,37 -----
verbose = selup & UP_VERBOSE;
quiet = selup & UP_QUIET;
if( sect.sct_lstup == 0 ) return(0);
! if( sect.sct_lstup < curup ) goto X160;
sect.sct_lstup = curup;
return(1);
X160:
***************
*** 197,203
amt = funit * q;
shipper = sect.sct_owned;
putsect(x, y);
! if( getsect(dx, dy, UP_NONE) == -1 ) goto X3404;
if( shipper == sect.sct_owned ) goto X3612;
if( sect.sct_chkpt != 0 ) goto X3612;
X3404:
--- 197,203 -----
amt = funit * q;
shipper = sect.sct_owned;
putsect(x, y);
! if( getsect(dx, dy, UP_TIME | UP_ALL) == -1 ) goto X3404;
if( shipper == sect.sct_owned ) goto X3612;
if( sect.sct_chkpt != 0 ) goto X3612;
X3404:
Jay Davis
jld@andrew.cmu.edutsf@drutx.ATT.COM (FisherTS) (04/24/87)
Be careful with those fixes to the rolling civilians bug. As Jay pointed out, the first attempt causes a recursion problem if two sectors deliver civvies to each other. This second attempt will follow a civilian delivery chain into an area where a hurricane could be lurking or plague might be starting. And if enough sectors are involved, I think the stack could be blown on some machines. I fixed this problem in Version 1.2 after experimenting with several ideas. It might not be perfect, but it doesn't seem to be wildly recursive or have unexpected side effects. I think the Version 1.2 fix involves adding a define to emp.h and a few lines in update.c. If you have 1.1, use the 1.1-1.2 conversion script to get the fix. In general, don't mess with update.c. It's fragile, and more than once I've been surprised by unexpected side effects. If you must change these sensitive routines, please rename your version to something other than 1.1 or 1.2 so I don't get it confused with something I sent out. Thanks. Tom Fisher ihnp4!drutx!tsf
wv@whuts.UUCP (04/30/87)
In article <3685@drutx.ATT.COM> tsf@drutx.ATT.COM (FisherTS) writes: >If you have 1.1, use the 1.1-1.2 conversion script >to get the fix. >Tom Fisher ihnp4!drutx!tsf Where can I get a copy of the conversion script? Is it short enough to post or email ? Bill Duncan ihnp4!mvuxi!wv
abc@mcnc.UUCP (Alan B. Clegg) (05/04/87)
In article <187@whuts.UUCP> wv@whuts.UUCP (54299-DUNCAN,W.) writes: >In article <3685@drutx.ATT.COM> tsf@drutx.ATT.COM (FisherTS) writes: >>If you have 1.1, use the 1.1-1.2 conversion script to get the fix. >Where can I get a copy of the conversion script? Is it short enough >to post or email ? The conversion upg1.1-1.2 is available via anonymous uucp and ftp from ncsuvx.ncsu.edu. If you are interested, please send me mail and I will give you information on dialing up. -abc
msj@gitpyr.UUCP (05/06/87)
In article <493@speedy.mcnc.UUCP> abc@speedy.UUCP (Alan B. Clegg) writes: >In article <187@whuts.UUCP> wv@whuts.UUCP (54299-DUNCAN,W.) writes: >>In article <3685@drutx.ATT.COM> tsf@drutx.ATT.COM (FisherTS) writes: >>>If you have 1.1, use the 1.1-1.2 conversion script to get the fix. >>Where can I get a copy of the conversion script? Is it short enough >>to post or email ? > >The conversion upg1.1-1.2 is available via anonymous uucp and ftp from >ncsuvx.ncsu.edu. If you are interested, please send me mail and I will >give you information on dialing up. Except that as of last night, the copy at ncsuvx is trashed. I ftp'd it to another system, uncompressed it and found the bottom half of it trashed. Mike /msj/ -- Mike St. Johns Georgia Insitute of Technology, Atlanta Georgia, 30332 ...!{akgua,allegra,amd,hplabs,ihnp4,seismo,ut-ngp}!gatech!gitpyr!msj StJohns@MIT-Multics.ARPA (404) 982-0035
abc@mcnc.UUCP (05/06/87)
In article <3539@gitpyr.gatech.EDU> msj@gitpyr.gatech.edu (Mike St. Johns) writes: ..>In article <493@speedy.mcnc.UUCP> abc@speedy.UUCP (Alan B. Clegg) writes: ..>>In article <187@whuts.UUCP> wv@whuts.UUCP (54299-DUNCAN,W.) writes: ..>>>In article <3685@drutx.ATT.COM> tsf@drutx.ATT.COM (FisherTS) writes: ..>>The conversion upg1.1-1.2 is available via anonymous uucp and ftp from ..>>ncsuvx.ncsu.edu. If you are interested, please send me mail and I will ..>>give you information on dialing up. ..> ..>Except that as of last night, the copy at ncsuvx is trashed. ..>I ftp'd it to another system, uncompressed it and ..>found the bottom half of it trashed. Mike I just checked the file and it is OK here (even the bottom). Anyway, please make sure you are in BINARY mode when you transfer the file, etc.. and contact me if you have problems instead of saying my copy is bad. Thanks, -abc