Carlos Manuel Duclos Vergara wrote: > On Friday 18 April 2008 16:05:56 Jeff Licquia wrote: >> Carlos Manuel Duclos Vergara wrote: >>> Now that I reread this, shouldn't it be the other way around? I >>> mean, if you are LSB 3.2 certified >>> for sure you are CGL, but if you are CGL then by definition you do >>> not fulfill LSB 3.2 requirements.... >> >> For *applications*, Dan had it right. An application that uses no >> more than the CGL subset interfaces will comply with the full LSB, >> since the CGL subset interfaces are all part of the full LSB. But >> an app certified against full LSB could, say, use interfaces from >> GTK+, which would not be part of the CGL subset. >> >> You are entirely correct, of course, when describing distributions. > > Well, I thought that we were talking about certifications for > distributions... after all the main reason this was proposed was to > certify distributions such as Montavista. Are we talking about > creating a special certification for ISVs? No, but for any available certification there are two pieces - distro and application certification. The CGL-ceritfied distro is a subset, so a CGL-certifie app would by definition also work on a full LSB system. Meanwhile, a full-LSB-certified app would probably include libraries that are not part of the CGL profile, and thus would likely not work on a CGL system. I think that's all anyone was saying. And this is the thing that is worrying some people: once you start segmenting the certifications, you have to start worrying about matching the right things together, and understanding the above matrix. I don't think this is a particularly severe problem in these controlled cases.