Em Quinta-feira 23 Setembro 2010, às 23:25:13, Greg KH escreveu: > On Thu, Sep 23, 2010 at 10:45:45PM +0200, Thiago Macieira wrote: > > On Thursday 23. September 2010 21.05.56 Greg KH wrote: > > > I don't see ConnMan providing an API or ABI, do you? If so, where is > > > it documented? > > > > Yes, I do. It has a D-Bus API. > > So, if _thats_ the requirement, then document that, not "you must use > connman".
Don't be naïve. The requirement is not the API, it's the whole thing. It's about looking, smelling and behaving like connmand. You may not like it, but the whole point is that the behaviour is provided all the way up from the low-level to the API. There will be companies/entities/people that test the stack and make their devices work with that stack. And there will be other companies/entities/people releasing software that depends on the behaviour of that stack. If you replace a core middleware component like connmand with something that emulates connmand, you risk having different behaviour. Sometimes for better, sometimes for worse. You wouldn't replace Qt with something that has the same API but a different implementation, would you? (no such thing exists, but see http://en.wikipedia.org/wiki/Harmony_(toolkit) ) Same for GStreamer. Or think too of glibc: there are alternatives, like uclibc and whatever Android uses. And again bash vs busybox. One of the drawbacks of the LSB is that it tries to mandate behaviour: it has a list of symbols and a battery of tests to validate behaviour. That way, distributors could simply replace one component with another that behaved the same. But you and I know that, as good as the LSB is, it's extremely slow- moving. Instead, MeeGo standardises by saying "here, you use this as a base, then you build on top". Just because connman is a small component and has a fairly self-contained API and predictable behaviour, it's not an exception to the rule. You must use the components, all of them, in the same versions. If an exception to connman is opened, then another will turn up for bash, one for uclibc, and so forth, to the point that it fragments the stack. -- Thiago Macieira - thiago (AT) macieira.info - thiago (AT) kde.org Senior Product Manager - Nokia, Qt Development Frameworks PGP/GPG: 0x6EF45358; fingerprint: E067 918B B660 DBD1 105C 966C 33F5 F005 6EF4 5358 Qt Developer Days 2010 - Munich Oct 11-13 - San Francisco Nov 1-3 For more information and to register: http://qt.nokia.com/qtdevdays2010
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ MeeGo-dev mailing list [email protected] http://lists.meego.com/listinfo/meego-dev
