[Starkit] corruption of starkits by double access

[Steve Blinkhorn]

>>>Jean-Claude Wippler said:
 > On Feb 27, 2006, at 13:30, Steve Blinkhorn wrote:
 > 
 > 
 > > One thing I *thought* might be raised would be the limitation of
 > > Metakit files to only one client for writing.   I half-expected
 > > someone to point to a non-fatal error that might be raised if a second
 > > client attempted to write to it, [...]
 > 
 > There is no error.  A second writer will ruthlessly damage the file,  
 > just as would happen with non-MK files.  The damage does not happen  
 > on open but on commit, BTW - so as long as you do no commit actual  
 > changes (and are effectively not a writer), nothing bad happens.
 > 

>From our experiments, building identical starkits write-enabled or
not, it seems that a starkit that is write-enabled is written to even
though no change has been made.   In other words some commit or other
happens, even when no change has been made, and this causes the
problem we started with.   Have I understood this correctly?

--
Steve Blinkhorn <steve at prd.co.uk>