Re: Is the eCos Open Source development model at risk?

[Thomas BINDER <Thomas dot Binder at frequentis dot com>]

Nick Garnett wrote:
> 
> <snip>
> I would certainly be willing to take a look. However I do have very
> deep reservations about including the sort of memory protection you
> appear to be suggesting. eCos was explicitly not designed to permit
> later addition of this sort of thing, and I would not want to
> compromise the current code or design to add it.

Nick!

As long as eCos was not deliberately designed to not support memory protection at all we are already (partly) happy :-).

To be more serious, the object-oriented way eCos was designed allowed us to incorporate many of the changes by simply inheriting from existing classes. That is, we were able to *add* most of the new parts, instead of changing existing code. However, there is a part that needed to be adapted, namely the hal. Although we certainly took care to make all changes revertable by a config option I know that this will probably be hard to verify for you, particularly for context.S and vector.S.

I could imagine it would make sense to either fork the hal, or to duplicate (some of) the changed files such that the CDL option merely selects which files (or which hal) actually gets compiled. I think I have read a post earlier this year in ecos-discuss addressing something similar.

Please let me know if you have any ideas on that.

> 
> If you want to just send me a diff against the current anoncvs then I
> will be happy to look it over. Don't bother about changelogs at
> present, unless they contain information that would be useful in
> understanding what you have done.
> 

Ok! I'll send the patch to ecos-patches.

best regards,
Tom
--