I'm trying to download a series of large files.  I enter 
the file names into the URL box, like:

<http://www.someweb.com/some/dir/large-file-01.mov>
<http://www.someweb.com/some/dir/large-file-02.mov>
<http://www.someweb.com/some/dir/large-file-03.mov>
etc.

But, due to a bad phone line or a flaky server or whatever, 
I'm only got some files whole, and parts of the rest.  The 
error log said something like "incorrect file size after 2 
retries" for most of the files.  So, I tried again using 
the "update" option.  While that was running, I reviewed 
the help files and found the "Number of retries" option.  I 
tried bumping that up from 2 to 999 - while it was running.

PROBLEM #1:  Altering the number of retries while mirroring 
is in progress doesn't seem to have an effect on anything 
except the text in the error log.  The program didn't take 
any longer than before, but still claimed to fail "after 
999 retries".

Next, I came here, found a new version, downloaded and 
installed it.  (I didn't think to take note of which old 
version I had been using.)  When I retried the download, it 
took nearly no time at all and gave no errors at all.  I 
found and activated the debug option, tried again, tracked 
down the debug file (which is NOT called winhttrack.txt 
like it says, but is actually hts-ioinfo.txt) and that 
seemed to say that all the files were current.

PROBLEM #2:  How can they be current if they're 
incomplete?  Does the thing verify ONLY against the cached 
info about the file and not the file itself?  I can 
understand this behavior for HTML files that are changed by 
HTTrack in the process of mirroring, but videos?
Finally, I deleted the partial files and ran it again.  
This seemed like an improvement, but still, if I download 
987,654 bytes of a 1,000,000 byte file before the Nth and 
last retry fails, why should I have to choose between 
living with a partial file or restarting from byte 1?
PROBLEM #3:  HTTrack only accepts a binary, yes or no 
answer to the question "was this file downloaded before?", 
when the real answer may well be "partially."  When rerun, 
the program should remember or detect incomplete files from 
the previous run and attempt another set of N retries for 
them - from where it left off last time, avoiding 
duplication of effort.