----- v1.10, 18 Mar 2004
+ new "b1increase" setting increases b1 by this much each time doecm loops.
+ doesn't "scientify" (1000000->1e6) reported b1 value if the result would 
  have an exponent less than 4.
+ "order = random" processes the composites in random order.
+ new "recommendedwork" setting let's doecm do the work that the server feels 
  is most needed. This only works if you're using the autodownload feature.
+ new "fallback" setting switches to "recommendedwork = yes" if your primary 
  list of composites is exhausted. Otherwise doecm quits when that happens.
+ new "method" setting allows easy switching between ECM, P-1, and P+1. Be 
  sure to coordinate your effort with the project coordinator if you intend to 
  do anything other than ECM though, as the others waste effort when several 
  people do the same thing. Just stick with ECM.
+ new "submitretryinterval" setting sets the number of minutes between retry 
  attempts when a factor submission has failed.


----- v1.01, 04 Mar 2004
+ fixed a minor annoyance in the cmdline parser.
+ if you had once gotten an autosubmit failure doecm would keep telling you 
  you had unsubmitted factors even though it had subsequently submitted them 
  properly.
+ included 11*2^n+-1 in doecm.ini.


----- v1.00, 03 Mar 2004
+ if there is some error during the autosubmit process the factor is now saved 
  to <submitfailurefile> and is regularly resubmitted until it succeeds.


----- v0.91, 03 Mar 2004
+ sends "args" and "submitter" fields when submitting factors.
+ downloads composites in gzipped format if "use_gzip = yes". This cuts 
  about 50% off the download size.
+ if "reportwork = yes" doecm sends a report of how many curves you ran, 
  what B1 you used, and what numbers you tried to factor once all composites 
  are processed. This will be helpful when guessing how big the remaining 
  factors are.
+ sorts composites in descending order if "order = desc".


----- v0.90, 29 Feb 2004
+ Initial release.