It could be an out of bounds read or write somewhere. Is it repeatable? (That is, assuming you made a backup of your world first, can you try running it again on a copy)
If you're up to it, you could install DrMingw (http://code.google.com/p/jrfonseca/wiki/DrMingw) as a debugger, and the next time the tool crashes, DrMingw would pop up with a complete backtrace at the time of crash. That would be very helpful to me, but don't feel pressured if you're not comfortable dealing with that stuff or you already have development tools set up that you don't want messed with.
Great, thanks for this. It's crashing in cnbt on read. It looks like malloc is failing. Interesting.
If you don't mind, I'd like to take further debugging to PM. It might require some back-and-forth to exactly pinpoint the problem. I'll send a new build in a little bit with some extra assertions to see if this is accurate.
Default mode without -oo or -ob will only overwrite rock with new ores? Or rock, cobblestone, gravel, and dirt? I'm about to run this on a heavily built SMP map and wanted to make sure I understand what it considers valid areas to place new deposits.
Thanks for the awesome tool - aside from updating for the Lapis, this will help us continue to use the same MP map without worrying about resource depletion.
The crashing reported by Pelvidar was a result of malformed chunks in the save, and the crash was originating in the NBT library as a result.
I have updated OreGen again to detect and report more severe instances of malformed chunks (which will also be skipped over, so that the utility doesn't crash). I would advise trying to verify these reports in another tool like MCEdit, and delete those chunks.
I ran across viewtopic.php?f=1022&t=127216 earlier today. I haven't tried it myself, but it looks very useful and if you have corruption bad enough that OreGen is reporting it, I would recommend running a more comprehensive checker like this one. You may have more subtle chunk corruption that OreGen is able to detect.
Topics about absence of lapis in pre 1.2 maps starts to emerge, and people are linking to your topic. For me command line software is no problem at all. I work on linux servers a lot so it's intuitive for me, but for many people it's hard to use it. Maybe you could write some kind of wrapper with GUI to your application. This way you can keep this tool the way it is and still let people without that knowledge to use it. McMap is created similar way and it works like a charm :smile.gif:
hmm, you mean just a gui where you have all options as widgets?
i could do this in qt4. it would only call the cli-program with the proper arguments however, since it would be my first c++-project and i’d like to keep things simple.
if i do it, i will publish the source, of course, so someone can wrap it up proplerly.
fileseletor: <world dir>
dropdown box: <ore_id>
optional (=preset) stuff:
spinbox: Number of rounds ore is generated per chunk
combined range slider: Minimum and maximum depth ore is generated (0-127)
range slider: Size of ore deposit (0-63) -- Defaults: Coal = 16, Iron/Gold = 8, Redstone/Diamond/Lapis = 7
disableable time spinbox: Only update chunks modified before time (as unix timestamp)
disableable time spinbox: Only update chunks modified after time (as unix timestamp)
checkbox: Take precedence over all existing ores
checkbox depending on the above one: Take precedence over all existing blocks (including air)
You could create a header file that contains relevant structs and function signatures. Or you could extern them. In theory there's only one function you'll ever directly invoke since everything is controlled by an options struct you pass through.