Re: [Sbcl-devel] CVS SBCL fails to compile on SBCL 0.9.12 if :sb-fluid is set

[Robert S. <st...@la...>]

Nikodemus Siivola writes:
 > > Reading the SBCL source code, I now understand that there must be
 > > something seriously wrong here (and not just a shortage of heap
 > > space).  I now realize that someone/something is actually trying to
 > > allocate an object that requires 340MB, while there is "only" 86MB
 > > *remaining*. 
 > 
 > Quite. One thing you could try is 
 > 
 >  (handler-bind ((storage-condition (c) (lambda (c) (sb-debug:backtrace))))
 >    ...load-gsharp...)
 > 
 > in order to see where the allocation is coming from.

Thanks for helping figure this out.  The problem was due to an object
with large amounts of shared structure being printed while
*print-circle* was nil.  That object is a Gsharp buffer that is shown
as part of an error message indicating that a particular slot is
unbound.  This error happens only when I start Gsharp from the CLIM
desktop and not when I start it from McCLIM+SLIME, and I now have to
figure out why.

-- 
Robert Strandh

---------------------------------------------------------------------
Greenspun's Tenth Rule of Programming: any sufficiently complicated C
or Fortran program contains an ad hoc informally-specified bug-ridden
slow implementation of half of Common Lisp.
---------------------------------------------------------------------