I tried to reproduce the crashing that Eugene reported, and could not. I have a decent C2D with 4 gigs of RAM, and with those specs all I could do was experience slightly-slower-than-expected performance when working with column-spanning headers. However, when I start my Ubuntu VM, and start the processes that cause it to consume about 2 gigs of RAM, then my system specs degrade to the minimum RAM acceptable to ID (just more than 2 gigs) and I start experiencing multi-second delays.

My hunch is that this particular feature requires more resources than the minimum spec from Adobe... or that a buggy plugin is crashing when it doesn't do garbage collection quick enough to prevent the host system from running out of available RAM. Just a hunch, I'm not a developer at all, but the whole thing smells like a memory leak to me. I could never get it to crash, though, so perhaps I'm only muddying the path with my posts about my attempts to reproduce it.

2 years on, and this issue still exists in CS6. Win7/64, 12GB RAM, delayed screen update.

I often - but not always - get it with split columns on a table. Other 'complications' may be column break character following, or an anchored object. I wish I had more time to isolate the issue...

I have tried exporting to idml, then reopening - same. The only way I can edit split table content sometimes is to remove the split, make the edit, then reapply the split 3/4.

Sometimes, even changing the size of a linked text frame that contains a split table is enough to lock up the program - even if there are no text flow changes from this, due to a column break character and plenty of free space below it...

And almost another year later and STILL no resolution?

Brand new, souped up macbook and crashes when I try to use span columns. Sometime, if there are only a handful of items that span columns, it works, though sluggishly. Adobe? Care to chime in on this one at all?

I think DeezleDeeDee was referencing several of the posters above that talk about minimum spec for InDesign vs possible spec required for this particular feature as a potential solution to the issue.  E.g., if a brand new MacBook can't cut it, and the bug was noted here 2.5+ years ago, then it's not (likely) an underspec-hardware problem as was contemplated above.