Is now posted here:
(link removed, fix now incorporated into Beta 3a - see the announcement for a download link for that)
We finally managed to reproduce the startup crashing bug that has been plaguing ipsi et al - it turns out that it only comes up when the software is NOT being run with a debugger, which is why we couldn't reproduce it before. And it did indeed have something to do with the radical database; there was a bug that in some cases could cause a variable in the radical table load function to be non-32-bit-word-aligned, which would result in a crash when it went into ARM processor mode. Where that variable ended up remained consistent on a specific system/build, which is why some people could start up the new beta without any problems and others would have it crash every time. Anyway, this should now be at least as stable as Beta 2 for everyone so once we get a little more feedback we'll put these fixes and the PPC ones into an official consolidated Beta 3(a) release.
(link removed, fix now incorporated into Beta 3a - see the announcement for a download link for that)
We finally managed to reproduce the startup crashing bug that has been plaguing ipsi et al - it turns out that it only comes up when the software is NOT being run with a debugger, which is why we couldn't reproduce it before. And it did indeed have something to do with the radical database; there was a bug that in some cases could cause a variable in the radical table load function to be non-32-bit-word-aligned, which would result in a crash when it went into ARM processor mode. Where that variable ended up remained consistent on a specific system/build, which is why some people could start up the new beta without any problems and others would have it crash every time. Anyway, this should now be at least as stable as Beta 2 for everyone so once we get a little more feedback we'll put these fixes and the PPC ones into an official consolidated Beta 3(a) release.