I definitely agree with you; languages should die, and they should die when their usefulness ends. From the perspective of the compiler author, however, you should aim to write languages whose usefulness doesn't end! :)
I look at it as they're are two types of languages - computer science languages which are for researching new concepts in computer science and software engineering languages which are used for building applications for commercial use and who's lifetimes may reach up to 20 or more years.
Languages in the first class should die quickly. They're intended to be for research. They tend to have a longer lifetime than they should when people use these languages for commercial use. Languages in the second class should be long-lived as the systems depending on them may be long lived. Some of the useful language features from the research languages may be introduced into the software engineering languages. C++ is a great example of this as it has been borrowing concepts from research languages for some time now.
taylodl|2 years ago
Languages in the first class should die quickly. They're intended to be for research. They tend to have a longer lifetime than they should when people use these languages for commercial use. Languages in the second class should be long-lived as the systems depending on them may be long lived. Some of the useful language features from the research languages may be introduced into the software engineering languages. C++ is a great example of this as it has been borrowing concepts from research languages for some time now.