Michel Dänzer wrote: >On Mon, 2002-04-08 at 13:45, Adam C Powell IV wrote: > >>I just received bug #141738 against petsc2.1.1-doc saying that links >>from /usr/share/doc/petsc2.1.1-doc/include are going to >>/lib/petscdir/2.1.1/include instead of /usr/lib/petscdir/2.1.1/include. >> >>This is really odd, as I made them using dh_link and the links seem >>correct (../../../.. should be /usr, right?), but indeed, they give file >>not found errors when I try to inspect the files. >> >>But what's *really* bizarre is: >> >>cd /usr/share/doc/petsc2.1.1-doc/include >>ls .. [gives what I'd expect, /usr/share/doc/petsc2.1.1-doc contents] >>ls ../.. [gives what I'd expect, /usr/share/doc contents] >>ls ../../.. [gives what I'd expect, /usr/share contents] >>ls ../../../.. gives not /usr, but root! >> >>What's wrong with the system, such that this happens -- for both me and >>the bug reporter? And why has this changed since I first made the >>symlinks and tested them, and they worked? >> >>I don't think this is a PETSc bug, but something deeper... >> >I can't reproduce this (but keep in mind that the .. entry in a >directory can theoretically point anywhere), but why don't you use >absolute link targets anyway? Looks like they would be shorter and less >error prone here. > I use dh_link, which gives the "correct" links as we're supposed to use them. I think relative links are specified in policy so that things will work right in a chroot, or something like that.
What do you mean by "the .. entry in a directory can theoretically point anywhere"? Okay, did a bit more poking around; on the machine in question: cd /usr/share && ls .. turned up the contents of root, but on other machines it did not. I think going to close this bug (141738) since it clearly has nothing to do with my package. Thanks, -- -Adam P. GPG fingerprint: D54D 1AEE B11C CE9B A02B C5DD 526F 01E8 564E E4B6 Welcome to the best software in the world today cafe! <http://lyre.mit.edu/%7Epowell/The_Best_Stuff_In_The_World_Today_Cafe.ogg> -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
