Hallo Martin, hallo Florian, now that we are discussing how to solve this bug, I'm adding it to the Cc.
martin f krafft <[EMAIL PROTECTED]> wrote: > also sprach Frank Küster <[EMAIL PROTECTED]> [2007.02.15.2002 +0000]: >> I have no problem with anyone fixing this. But I doubt that ucf >> is needed and advisable here. Why not just parse and edit the >> configuration file? > > Because parsing configuration files is a major pain and very > error-prone? That depends on the format of the configuration file, see also the other mails in the bug log. >> If you do use ucf, please don't forget ucfr. > > I won't forget it. However, I think ucf is the way to go. ucf is something a maintainer may like or dislike. Personally, I don't like it very much, but there are cases where there is no good alternative (there were more such cases in sarge than in etch). The problems with ucf are: - While it does some things much better than dpkg, it does only half of dpkg's job (see ucfr...) - If it isn't strictly needed, it only adds complexity - It isn't needed here, none of the cases is involved where it performs better than dpkg (- ucf has already been advocated too much. People think using ucf on a file is enough to comply with policy requirements, and produce bogus "file changed by you" prompts.) Regards, Frank -- Dr. Frank Küster Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich Debian Developer (teTeX/TeXLive)
