Hi, I tried to do this: %waste2 is just about 3 megs of junk data. I want to build %waste, a file with about a gigabyte of junk. x: read %waste2 o: open/binary/direct %waste for i 1 300 1 [ insert tail o x ] close o when I do this, the file doesn't usually grow. If it does, it always starts from the beginning as if the "tail o" part were ignored. Is this just how /direct works? Why can't I seek to the end of the file and just append some data without some buffering getting in the way? If I don't say direct then rebol tries to buffer the whole file as a series, and needless to say crashes or says out of mem, etc. on large data like this. I finally discovered write/append but why can't I work with the file as a series properly? My OS is NT and it certainly has random access to any part of the file, just as any part of the series, so you should have no trouble mapping the rebol series operations to the OS file operations. Has anybody reported this problem? I discovered it using /Core, have not tested /View yet. Thanks, -galt