Create two files:

  aaa.txt - hundreds of MB
  bbb.txt - a few bytes

Select them both, and copy them to another directory.

You'll get a file progress window.  The files get copied in reverse alphabetical order (or something like that).  If bbb.txt gets copied *first*, and before the progress dialog comes up, you'll see that the dialog says "copying bbb.txt, file 1 of 2" when it is obviously copying the large aaa.txt instead.

By adding some printf()s to handle_transfer_ok(), we get this:

PHASE_INITIAL
NO PHASE (DEFAULT)
PHASE_COLLECTING
PHASE_COLLECTING
PHASE_READYTOGO
PHASE_MOVING/OPENSOURCE/OPENTARGET/COPYING:
  source_name: file:///home/federico/Desktop/nld10/bbb
  file_index: 1
  bytes_copied == 0; setting full text
PHASE_MOVING/OPENSOURCE/OPENTARGET/COPYING:       *******
  source_name: file:///home/federico/Desktop/nld10/aaa
  file_index: 2
  bytes_copied == 126976; just setting sizes
PHASE_MOVING/OPENSOURCE/OPENTARGET/COPYING:
  source_name: file:///home/federico/Desktop/nld10/aaa
  file_index: 2
  bytes_copied == 524288; just setting sizes

See the line marked with *******.  We never get a "bytes_copied == 0" for the large file, since part of it already got copied before the dialog came up, so we never get to updating the labels.

I don't think it hurts to just change the labels *and* the sizes every time; there is no need to make the logic more complicated by special-casing the first 0 bytes of each file.