Babcock: (from Reprogramming)
»We see this problem divided into three areas of discussion
1. Conversion
2. Re-writing of systems and programs
3. Future trends
Most users have faced the conversion problems before. The cost is usually a function of the severity of the conversion, that is, from one machine to another (usually highest cost), from one operating system to another and from one language to another. The most effective means today (but not the most desired) is by rise of hardware techniques, that is, emulation.
New systems usually exhibit features that are desirable and that handle a broader range of application in a more efficient manner. To utilize these new tools, re-writing is sometimes dictated. This is perhaps the most costly of the three areas but is mandatory in many areas. We see the need for higher levels of languages, such as compiler languages of compilers, in order to make this decision less critical.
In the future, we see the need for both hardware and software facilities specially designed for the overall problem of transporting. The stress will be as great on hardware design as on software engineering because the key to portability can be exhibited in the near future whereas software engineering has not yet produced nor announced a true language of languages for hardware independence. Hardware portability functions can be economic in present day ‘Assembler-oriented’ applications, but we hope for integrated hardware/software facilities to reduce substantially the costs of the reprogramming problem.«