Cart
[Home] [By Thread] [By Date] [Recent Entries]
> "Henry S. Thompson" wrote: > > There's a widely-supported assertion in computer science that > > text-substitution macros are dangerous mechanisms to use to bring > > extensibility to a language. Yes, but it should not be over-rated. When dealing with variants and time, many systems end up coming back to text substitution: if the language does not provide ways to do it then some source code control system ends up providing it by magic (internal storage of diffs, or other entity-level issues). There is some complexity that it is fruitful to model with general abstractions, and other complexity is rather arbitrary and is better handled by a low-level mechanism that at least allows the Job to get done. The common phenomenon in C++ was that people continued to use #ifdef to handle variants (such as different localized versions, or versions for different platforms) even when using objects: by partitioning off difficult issues into a low-level mechanism they made cleaner use of the abstraction mechanism. Rick Jelliffe
|
