Hi, On Monday 14 March 2005 00:23, Aatu Koskensilta wrote: > Hello. > > I've been studying DirectFB for a few days now, and am going to adopt > it for a small digital camera project I'm involved in. Since I'm going > to write the required software in C++, I took a look at snapshots of > the C++ wrappers ++dfb and dfb++ that exist. Neither of these met my > subjective quality criteria, e.g. since a plain C function pointer is > passed to the enumeration functions instead of a type safe > boost::function or such like. > Also, most of the helper data structures > were not wrapped in C++ish helper classes, which really should be done. > Finally, there really is no need for the macro constructions used as a > simple template suffices. Because of these reasons I chose to write a > C++ wrapper library for DirectFB myself, which turned out to be a > rather trivial although somewhat tedious task. >
depending on your point of view - how to wrap c functions - of course you can feel free to implement your own wrapper functions and classes. of course you are able to wrap it with lib boost components. now, from my point of view: directfb was @first designed to have a fast and memfriendly framebuffer application interface. using directfb on a pc is more than a positive sideeffect, but think about directfb's first target hardware - embedded devices - e.g. set top boxes. those don't have nowadays resources like gigabytes of ram, harddisks ect. you have to have your application - including linux - to fit into e.g. 8 mb of flash ram. also working memory doesn't exceed 32 or 64 mb normally. and there is the next fact - the processor. on a stb you get a kind of embedded cpu's like mips, powerpc, ect. those are running from 66MHz to >250MHz. those reasons are responsible for having ++DFB and DFB++ as small wrappers. get usage of c++, but keep on small size and fast execution. > However, complaining about the state of the C++ wrappers is really not > my point. Rather, I'd wish to know whether people are happy with the > current C++ wrappers, or whether my disappointment is shared by others. > If the latter is the case I could put in the effort to wrap *all* of > DirectFB in nice C++ helper classes instead of just the portion I need > for my project. (The whole code is going to be released under GPL or > LPGL in any case, so my question actually is whether people think that > a new "official" C++ wrapper for DirectFB would be a good thing or > not.) as mentioned before - feel free to implement a ++DFB-NG or whatever you will call it. i'm looking forward to see it appearing at cvs-ml :) > > Awaiting opinions, Regards Marcel _______________________________________________ directfb-dev mailing list [email protected] http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-dev
