<URL: http://bugs.freeciv.org/Ticket/Display.html?id=39563 >

Found the old report and merged with it.  Added some PR# links, instead 
of the revision numbers Marko used in the report.

Funny, I've been waiting for Marko to fix this, as it's his original 
report, but now I see that Per flagged it for me!  Since Per never 
sent me an email, I'd no idea....  And I've been too busy fixing the 
myriad 2.1 crashers to worry about non-crashing log messages like this!

As Jason notes, the problem is that Per set the log level to LOG_ERROR, 
but the comments say this is OK....

ai/aicity.c: ai_city_choose_build()

  /* Fallbacks */
  if (pcity->ai.choice.want == 0) {
    /* Fallbacks do happen with techlevel 0, which is now default. -- Per */
    CITY_LOG(LOG_ERROR, pcity, "Falling back - didn't want to build
             " workers, caravans, settlers, or buildings!");
    pcity->ai.choice.want = 1;

However, I'll note that this was repoted after I calloc'd the city 
(instead of malloc).  Before I touched it, nobody ever cleared the 
ai.choice, so want (and boat) had garbage in them.

(I discovered the ai.founder_boat problem working on PR#39365 in May, 
that led to finding more and more uninitialized ai variables.)

If you look closely, I've even documented the former lack of 
initialization in common/city.c: create_city_virtual()

  /* pcity->ai.choice; placeholder for searching */
  pcity->ai.founder_boat = FALSE;

All in all, though, I'm not really the person to go hunting for the 
ai.choice.want problem.  Per would be the expert maintainer, and 
Marko was the original reporter.  All I've done is initialize it.

Freeciv-dev mailing list

Reply via email to