On Thu, 30 Jun 2005, Jon Ericson wrote: > Kevin Atkinson <[EMAIL PROTECTED]> writes: > > > On Wed, 29 Jun 2005, Gary Setter wrote: > > > >> In version 0.60.3 module suggest.cpp function > >> Working::try_scan(), the following line was removed: > >> assert(strlen(sw->word) == sw->word_size); > >> > >> I'm sure it is a good change, but would anyone care to mention > >> why the change? > > > > Performance. The condition still must always be true. > > Why not define NDEBUG at compile time instead? The whole point of > assert is to have no performance impact when making tests that must be > true.
I do not believe in ever disabling asserts. Often an assert can catch a bug earlier on before it develops into a serious problem. I do not wish to argue this point. _______________________________________________ Aspell-devel mailing list Aspell-devel@gnu.org http://lists.gnu.org/mailman/listinfo/aspell-devel
