odd error trying to hack FastString implementation

Isaac Dupree isaacdupree at charter.net
Mon Aug 13 11:14:12 EDT 2007


Currently FastString uses a custom-made hash table for their
construction.  I wanted to see how performance compared, using a binary
search tree instead.  Using FiniteMap made a module import loop that was
hard to resolve, so I switched to Data.Map for testing (I know it's only
available since ghc 6.4).  ghc-6.6.1 compiled stage1 okay and then
'base' was compiled okay, but then in 'containers', Data.Sequence
errored about problems with other interface files (different errors the
two times I tried it).  I couldn't figure out what was going on;
--show-iface and diff for
libraries/base/dist/build/Text/ParserCombinators/ReadP.hi between
working (unpatched) (well, except tc095(normal) failed), and the version
with my hacky patches, only showed
-interface base:Text.ParserCombinators.ReadP 1 60720070812 where
+interface base:Text.ParserCombinators.ReadP 1 60720070813 where

Any guesses what's going on here? or what more information might be
helpful, or where to look?

Isaac



More information about the Cvs-ghc mailing list