SHahn@SUMEX-AIM.ARPA (03/17/84)
From: Sam Hahn <SHahn@SUMEX-AIM.ARPA> Thanks for your reply. In summary, there are several ways: 1. Use DDT to break, and PIP to rebuild, the large files. This is tedious and error-prone, but will work. 2. Use PIP with the start and end options. This requires knowing a lot about the insides of a (possibly binary) file. Also, the segment sizes are not easily determined. 3. Use SQ and USQ to Huffman-ize the files. For me, there was quite a savings in space to SQ my file, and this worked. Does have a limitation if your binary file doesn't have a good distribution of char's for SQ to really make a difference. Also, still have a problem if the squeezed file is still too big. SQ'ing a second time still reduces, but there's a limit. 4. Use a HEXIFY program. This is a pretty good solution which requires a step to break up the hexified program. If the source file were nearly the size of a disk, one is pretty stuck, since the hexified file is larger than the original; probably want a hexify program which can operate in stages (could be this already exists -- in the end, I didn't need to do this). 5. Custom-build a program which steps through the original file, and copies onto another disk until that disk is full, closes, resets, and continues until N disks are full and the original file is scanned. Most time-consuming solution, but there will be cases in which this is the best solution, I imagine. 6. Do directory twiddling to rename the extents of the big file so that each extent looks like a separate file. Would require directory munging of the file, so potentially dangerous, but probably doable in an easy and "correct" way so that the directory is not damaged. PIP could than be used to rebuild the file. Thanks to all of you who responded. For those of you who didn't see my original request, I needed to transfer large (>400k) files from a non-standard (SDSystems) hidensity format to my new Compupro hardware. My apologies for running on at such length. -- sam hahn -------