Ross, I changed the query to "count(distinct package)" and indeed the "noises" of same package imports are filtered out. The two top-10 lists now look somewhat similar. GHC only: http://haskell.ecoin.net/cgi-bin/modules.pl?src=ghc 1 Control.Monad 22 2 Data.Char 20 3 System.IO 17 4 Data.List 16 5 Data.Maybe 16 6 System.IO.Unsafe 14 7 Control.Exception 13 8 Foreign.Ptr 11 9 Data.Word 11 10 Foreign.C 11 GHC+HackageDB: http://haskell.ecoin.net/cgi-bin/modules.pl?src=both 1 Control.Monad 178 2 Data.List 132 3 Data.Maybe 113 4 Data.Char 104 5 System.IO 99 6 Control.Exception 98 7 Data.Word 83 8 System.IO.Unsafe 81 9 Foreign 74 10 Data.Map 68 Steve On Sat, Feb 23, 2008 at 5:25 PM, Ross Paterson <ross@soi.city.ac.uk> wrote:
On Sat, Feb 23, 2008 at 10:59:40AM -0500, Steve Lihn wrote:
I parsed through all the hackagedb modules. I also added the display of repository source (ghc, hdb) and package source. HackageDB is 3-4 times bigger than GHC core. The result is interesting, looking at the most used modules move up and down the rank...
You'd get a different picture again if you excluded same-package imports. Some of the high-ranking modules are used internally in large packages.