Roland Rosenfeld <[EMAIL PROTECTED]> wrote: > reassign 404662 iceweasel 2.0.0.1+dfsg-1 > retitle 404662 iceweasel: creates /usr/share/firefox symlink > severity 404662 important > thanks > >> Unpacking replacement iceweasel ... >> dpkg: error processing >> /var/cache/apt/archives/iceweasel_2.0.0.1+dfsg-1_i386.deb (--unpack): >> trying to overwrite `/usr/share/firefox', which is also in package dwww >> Please restart any running Iceweasels, or you will experience problems. >> >> I guess that directory belongs rather to iceweasel than to dwww. Hence the >> bug belongs to your package. > > The problem is that /usr/share/firefox usually is a directory (at > least in the packages dwww and firefox), while iceweasel contains a > symlink /usr/share/firefox:
I wonder why this causes a problem in dpkg at all. ,---- Policy 6.6 | 4. [...] | | A directory will never be replaced by a symbolic link to a | directory or vice versa; instead, the existing state (symlink or | not) will be left alone and dpkg will follow the symlink if | there is one. | `---- To me this always implied that file conflicts between packages aren't possible in this situation, either (and I even think I tried it out). After all, it's not a bug for two packages to ship the same directory. Moreover, I don't see how this problem can be fixed in iceweasel, and why it is not RC for dwww to fail to install when iceweasel is present. One of the changes from sarge to etch is the switch from firefox to iceweasel, and packages that want to cooperate with this browser have to follow that change. > Maybe iceweasel should conflict old/current versions of dwww, if it > changes /usr/share/firefox from a directory to a symlink? > > At least this isn't a bug in dwww 1.10.0, because its behavior in etch > is correct since firefox 1.5.dfsg+1.5.0.7-2 is part of etch and this > package contains /usr/share/firefox as a directory, not as a symlink. AFAIK etch will not be released with firefox - both because of the trademark/freeness issues that led to the iceweasel fork in the first place, and because I don't think the security team is willing to support this old version, when a new one is available. Regards, Frank -- Dr. Frank Küster Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich Debian Developer (teTeX/TeXLive)
