0936
   How to write the sum files to make checking easy?
   So, data comes in.  Sum file goes parallel to other data leaves.  We
   could give it the same name if other sums had a name that always sorted
   after or separately.
   We could give it a different extension.  Like .sum and .sums or such.
   My autocorrect turned sumsum into sums.  Not often I accept what it
   does.  Sometimes it has me capitalise words in the middle of sentences
   and such.
   0940 using .check for single files
   So, then we can sum all the .check files into a further file I suppose.
   0942
   Ok um
   So
   Um
   ./check *.checks *.check # seems to work fine
   When making new .checks files, increment a counter by 1 or such
   Maybe we can use date
   Date works
   Ok when making new check files, in theory we only need the last created
   one, and the one just made.
   0946
   My scripts could be simplified if one read from stdin instead of a
   file.
   0948
   Oops!  It seeks to 0 repeatedly.  That wasn't the right idea.
   0951
   Ok the system is busy testing stuff.
   1015
   1019
   Okay, I'm testing a little now.  Hashing a lot of data takes ages on a
   raspberry pi.  The current script is written to do all the hashes and
   move forward when they are done, but the user really would want some
   string to verify with before it is done.
   With the current file approach, that could mean a separate file for the
   first hash, or each hash.
   It might be more organised to hash based on lines rather than
   filenames.
   Meanwhile, my lotus client finished a task.
   1020
   ERROR: generating unsealed CID: Piece is larger than sector.
   So, filecoin needs data in 32GiB chunks at the largest.  I can delete
   my 64GiB test files.
   1021