[ECOS] Re: C++ kernel API

Bart Veer bartv@cygnus.co.uk
Sat Dec 5 10:56:00 GMT 1998


>>>>> "Stuart" == Stuart Adams <sja@brightstareng.com> writes:

    Stuart> Is there an plan to support/document a C++ kernel API ???
    Stuart> (Since the underlying implementation is C++ it should
    Stuart> simple right ?)

Actually, if you want to do things properly then it is usually more
difficult to publish a C++ API then a C one. Depending on exactly how
you set about doing it, for example how much functionality will be
available in the form of inline functions, a C++ API tends to expose
much more of the implementation details than an equivalent C API. Such
implementation details then have the unfortunate habit of sneaking
into application code. As a result it would be more difficult for us
to move the system forwards without breaking existing code.

There are certainly plans for a C++ kernel API, but we want to get it
right. Or at least as close to "right" as we can make it. People who
want to program entirely in C++ right now can do so by looking at the
header files - usually there will be an obvious correspondence between
the C++ classes and the published C API - but there is an increased
risk that the application code will need to be updated as the system
evolves.

Bart Veer // eCos net maintainer


More information about the Ecos-discuss mailing list