On 02/07/07, Alberto Ruiz <[EMAIL PROTECTED]> wrote: > 2007/7/2, Shawn Walker <[EMAIL PROTECTED]>: > > If package naming is really that much of an issue, then I could > > probably sit here all day and show you the rather convoluted and > > unhelpful names of packages I find on many GNU/Linux distributions > > most of the day. > > Forget about GNU/Linux, I don't care if the GNU/Linux distributions does > well or bad packaging naming. I do care about the one we are going to use. > Don't turn this into a useless fight between GNU/Linux way and Solaris way. > This is about good design principles.
This is not a fight between GNU/Linux way and Solaris way, though I could see how you might believe I implied that. It seems logical to use GNU/Linux as a point of comparison since this whole project is about making things friendlier to GNU/Linux users. "If I wanted a fight, I'd fightin' words," as my grandpappy used to say ;) One of the good design principles behind the current naming system for Solaris packages is that by prefixing a package name with a unique identifier (such as SUNW) there is no need for a central package registry to prevent package naming conflicts. As an example, there are *many* programs out there called allegro and even a few libraries. In a truly distributed system, you can't rely on unique package naming without some sort of consistent unique prefixing. It is the same reason that Java modules have the com.sun.* and so forth in their names. Let's say I want to package two different programs that are both called allegro (unfortunately), but they both install to completely different locations or provide completely different functionality. Obviously, they both can't be called allegro, but, you could call one SUNWallegro and the other FOOallegro. Of course, if Sun is the provider of both, you're in trouble, but I digress ;) > > I do not believe that a tool cannot be provided to make it easy to > > find and manage software that does not rely solely on arbitrarily > > chosen, and not always helpful, package names. > > Okay, then write it, and show me that I'm wrong. Meanwhile, even the better > tools for package managing out there, needs dealing with package directly at > some point. I never said they didn't. I just merely pointed out that dealing with the package name alone is not adequate. That's why rpm (as an example) provides an easy way to find what package a file belongs to using -q --whatprovides options. > On the other hand, we are forgetting, that this is a community effort, and > we are going to expect community members to maintain packages. Keeping good > principles and conventions within package naming, will make life way easier > to package maintainers. I don't think anyone is forgetting this is a community effort. The current naming standards for Solaris were chosen so 3rd parties could work together with Sun with less problems. But part of a community effort is realising that, so far, the existing community has overwhelmingly indicated how much they value backwards compatibility. > > > Indiana should be designed for the long term, as good as possible. Of > > > course, compatibility with Solaris should be kept whenever possible, but > it > > > shouldn't stop meaningful problems. > > > > In the grand scheme of things, I sincerely doubt package names are > > going to be what stops adoption of any OpenSolaris based distribution. > > I think there are many other things that are far more meaningful in > > the long run. > > At this point, I think its pointless trying to discuss, again, the > importance of a sane packaging system and the use of sane conventions in it. > I wont convince you on how important it is, but that doesn't change the fact > that it is extremely important for a lot of people. That would imply the current conventions are not sane, which is not true. The current conventions were not chose arbitrarily, but rather with purpose and great reasoning. -- Shawn Walker, Software and Systems Analyst [EMAIL PROTECTED] - http://binarycrusader.blogspot.com/ "Beware of bugs in the above code; I have only proved it correct, not tried it. " --Donald Knuth _______________________________________________ indiana-discuss mailing list [email protected] http://opensolaris.org/mailman/listinfo/indiana-discuss
