On Fri, 2004-12-17 at 19:35 -0500, Jeff Spaleta wrote: > On Sat, 18 Dec 2004 01:18:36 +0100, nodata <fedora@xxxxxxxxxxxx> wrote: > > I parse filenames for both version and repository information. > > Are you an end-user facing tool? I'm not talking about scripts you > maintain for yourself. I'm talking about tools that are produced for > other people to use. I'm pretty sure a lot of people do a lot of > clever things with script logic that is fragile and not behavior to be > encouraged or relied on. For as much as we want to all believe > distags are a standard process...it isn't. Its a hack to work around > default settings in the rpmbuild setup. And not all packagers are > using that hack yet either... some do...some don't. Whether or not > having a disttag in the filename is not the issue. The issue is > polluting a release tag with non-comparable information because its > the quick and easy thing to do. > > And I'm not talking about scripts used inside a buildsystem where the > build policy is clearly laid out to use disttag consistently.... > whatever buildsystem that is using disttags consistently right now in > its build scripts to parse filenames can get the same behavior from > using a header tag and placing the header tag in the filename and > still get the same filename parsing. This information does not belong > in the release tag which is used in version comparisons by librpm. > Not being able to decided how to use an existing tag to keep this > information seperate is a copout. Charles here has given you the > exact example on how to use a seperate tag to encode the exact same > filename without polluting the release tag. > The popular solution is a burden to the version-release comparison > process and it needs to be fixed. > > -jef"what is right isn't always popular and what is popular isn't > always right"spaleta > > If I download an rpm from anywhere and install it, I keep a copy of it. Later, if I want to use any of the standard *nix commands to deal with those rpms, I can do. For example, an ls *.dag.* shows me dag's rpms. Not having an important piece of information like "where the rpm came from" immediately seems a bit silly. Sure rpmlib can get at the information, but that doesn't mean it shouldn't be in the filename. People use filenames. Libraries can parse the files.