Re: [Haskell-cafe] Reading integers

Hi, I've commited some small improvements to the readInteger code, which speed up handling negative numbers and also the compatability wrappers very slightly. My last numbers comparing the reads wrappers to the basic readInteger code turned out to be unfair; apparently ghc managed to share some of the calculations of these functions somehow. I've changed the benchmark code to wrap the new functions into NOINLINE helpers, so that source of errors is eliminated. Now the reads overhead looks much less scary then before, on the order of 10% for small integers. regards, Bertram