From the ChangeLog:
- Added a new constraint, activities occupy max rooms (which can, for example, constrain some activities to be in the same room (in different time slots, of course) if max occupied = 1). Suggested by Yush Yuen, nomad, Volker Dirr and maybe others.
- Safe saving of .fet files (save with .tmp name, then remove old .fet file, then rename .tmp to .fet).
- Added an Uzbek translation, by Orzubek Eraliyev.
Please help with testing the new constraint.
Link, as usual, http://lalescu.ro/liviu/fet/download/test/ (only tar.bz2 for now).
I added also a Windows executable.
Translators, there are some new fields in the FetMainForm_template, please translate these, at least.
I added a new snapshot.
By Volker's suggestion, the new constraint contains also the word "different": Constraint activities occupy max different rooms.
Also, I corrected some bugs in this new constraint's dialogs.
To use with your old files, you need to replace globally "ActivitiesOccupyMaxRooms" with "ActivitiesOccupyMaxDifferentRooms" and "Max_Number_of_Occupied_Rooms" with "Max_Number_of_Different_Rooms" in your .fet file. Sorry!
Link, as usual: http://lalescu.ro/liviu/fet/download/test/
yush reported a problem, which is confirmed. See attached files (made by me). If the user adds all preferred rooms to all activities, FET cannot solve the timetable (FET has too many choices).
I'll try to solve. But it seems very hard now.
So, either:
1) I'll find a perfect solution (which will be slower than current solution, because of additional tests).
2) I'll mark the new constraint as "Not perfect", or really "Dangerous! Use only in case of strict necessity!" and treat similarly to activity tag max hours daily and students max gaps per day (but it is really more dangerous!)
3) I won't add it to official FET (I hate this, and I think all of you do).
Problem solved optimally! :-)
See http://lalescu.ro/liviu/fet/forum/index.php?topic=1136.0 . I solved an old incredible bug and the files are solved (well, quite slow, but I cannot get better).
I put a new snapshot, same link, http://lalescu.ro/liviu/fet/download/test/ . Please test on your files, because generation now is different (I mean the time and random seeds). Your files should solve in comparable time.
I tested on all my important files files, everything is good, times are comparable (no speed improvement on old constraints, the only improvement is on the new constraint, in which old buggy version gives impossible timetables), so I am thinking of releasing the new 5.18.0 version in a week from now.
Translators, please send your updates before the release.
I added a new snapshot, minor (or major? But I don't think so) improvement in the new constraint. Again, discovered by yush's report.
I added a new snapshot, it is more like a theoretical correction, but might have practical importance. Again, suggested by yush's reports. The random seeds again change. I hope it's the last change :-) Sorry!
I added a new snapshot, minor improvement, now the random seed after the generation is also written.