Agreed, this message is merely an information, maybe it could be suppressed.
I did trace and count the occurrences during one game, and 99% of times it needs less than 1800 iteration, the distribution is decreasing very quickly ( something like : probability ( N ) = O (1 / sqrt N ) )
And when the situation in one city starts to fail in a reasonable number of iteration, it will continue to fail (city is too big , or has too many costs ...)
Agreed , this message is just an information, it is normal to get this message from times to times
I counted during one game, and the distribution decreases very quickly, something like prob (N_iteration) = 1/ sqrt (N_iteration)
And when it starts to need a lot of iterations, this is due to the city (too big, or too many units or costs ...), so no need to increase the limit (i would even advocate to lower it to 5000).
On my "testmatic" setup I have to keep just one local patch any more. Would want to make also that one part of freeciv repository to ease things (no forgetting applying the local patch, no conflicts in applying it after repo changes, not getting it accidentally included in diffs...)
It's about turning the message "Did not find a cm solution..." to log_debug() one. It's a message I don't want to fill the error logs. It's not *that* serious, it happens a lot, and I've got otherwise clean logs (i.e. I can treat any appearance of log as a more or less critical situation to check)