On Saturday 01 March 2008 20:23:08 Bob Rogers wrote: > r26053 produces the segfault shown below, which looks like another > case of trying to compare the HLL name "lisp" to something corrupted > from the HLL name-to-ID hash. (FWIW, it was still broken as of r26165.) > > However, if I revert the src/pmc/string.pmc change, it works again, > and also in r26175.
How's r26170? I expect that to work and r26169 to fail. That is, hopefully the changes I made to HLL storage in r26170 squashed this for good, at least until we get our constant/GC problems solved. > The failure happens about 2/3 of the way through loading bytecode > files at Kea-CL startup time, so I could create a test case if you > wanted to poke at it yourself, but it would be pretty big. WDYT? Do you have any dynPMCs, especially of the String kind? If so, do they map to any internal PMCs? -- c
