ISO recognizes MISRA and the EU meets China

1/21/2010 11:13 AM EST

There are also standards for safety. These are required so that the industry has guidelines to work to without having to re-invent processes, methods and liabilities every time, thus saving money (and lives).

In the late 1980s, despite the K&R book on C, the software industry (and compiler makers in particular) needed a Standard for C. Therefore, it was done quickly and by the industrial players. After that, many other people got involved and C99 had an input from many people, not just the tool makers and those with a financial interest in a standardized language. The result was that C99 contained a lot that a few people thought was "cool," but the industrial players were not interested in. Had they been, the compiler makers would have implemented C99 in months rather than all but ignoring it for a decade. Note the // for comments was implemented before it was in the Standard.

In BSI, anyone can join the BSI language panels—you just walk in. Apart from the industrial players, many others did who were "joe programmer" and also some who, whilst not actually having much in the way of qualifications or experience, wanted to do something in standards. It was almost their hobby. You can meet people like this in many clubs and societies where it can be a substitute for a real career.

This caused some problems in the BSI C panel in particular. Because ISO Standard C compilers are not required by law or any commercial imperative (and for the embedded world, we were quite happy with C90/95), the industrial players were less interested in putting lots of effort into a panel that was messing about. Therefore, the BSI C panel in particular became lopsided. There was a lot of infighting and BSI suspended the C panel two or three times due to the behaviour of some members. Eventually, about a year ago, it was closed rather than suspended. The problem is that many of the causes migrated to the panel above. So should BSI ever re-open the C panel, it will be in the same mess as when it closed. A new broom is required to sweep out the BSI languages group.

On the other hand, MISRA-C, C++, Auto-Code, Safety Analysis, and Languages have small teams of qualified and experienced people drawn from industry because the industry wanted the MISRA standard. Initially, MISRA was for the automotive industry. Now it serves aerospace, automotive, and industrial and tool vendors.

By the way, the tool vendors watch each other and the rest of us watch them, so there is no bias. Someone commented that MISRA-C2 was MISRA-C1 for tool vendors. Well, yes and no. Tools work to rules. Mechanically, they don't use fuzzy logic or guesswork. The tool vendors, in general, not just the ones on the MISRA panels, were asking for clarifications on ambiguities in some of the rules. Thus, by clearing up the ambiguities, the tool vendors were able to implement more of the rules. Then again, so was everyone else.