Hyrum K Wright wrote on Mon, May 16, 2011 at 10:05:50 +0000: > It turns out that on trunk, we are using apr_int64_t which Daniel > tells me is still incorrect,
For the record, Roderich pointed it out too: > >> Patch below. The problem is still present in branches/1.6.x, while trunk > >> has > >> rewritten packed_offset_cache() and shard is now an apr_int64_t (which > >> will work, > >> though still not literally consistent with the creation of the cache).
