2005年10月4日火曜日
Naive implementation and user-friendly implementation
While reading codes of NZMATH, I've noticed there are many cluttering type checks and such for user-frienliness. For speed up in the future, they seem to be going to become an obstacles; it is better to asuume the validity of inputs. In other words, 'implement naively.' On the other hand, the user-frienliness was introduced for users' (or developers' as well) convinience, and it is not negligible factor.
2005年3月16日水曜日
No More New Toy Languages
Mathematical systems often provide their own programming languages. Most of them are, however, toy languages. They lack some useful features that ordinary programming languages surely provide. For example, SIMATH front end doesn't provide the way to define a new function. Another example is that gp of PARI had no local variables.
Nowadays, there are several languages with
- an interpreter and
- object-oriented features,
such as Python, Ruby, etc.. We guess they are good enough to be used as the interpreter of a mathematical system. Thus, NZMATH is provided as a library for Python. Users can write their programs with a true language.
登録:
投稿 (Atom)
