Angus Leeming wrote: > Georg Baum wrote: >>> Surely, argv[0] is by definition either an absolute path or is relative >>> to CWD. The shell expands everything else. No? >> >> I am not sure. Maybe there are some weird shells or OSes that behave >> differently? Then points 1. and 2. make sense. Or the author of it did >> simply not know that he does not need to expand the path manually. > > Indeed, that would make sense. However, this code is from os_unix.C. It > makes no sense at all for a unix shell, does it?
AFAIK no, but you are the expert here. > I ask because ExpandPath requires that os::homepath() be set. Since these > paths are exactly what is being set in this body of code, it suggests that > the result will be fragile to changes in the order of which os::homepath() > and the "absolute path to the lyx executable" are set. Then I would suggest to remove this stuff unless somebody remembers why it is there. BTW, while you are at it: Could you change the code so that calling lyx-qt, lyx-xforms etc. directly works? It is rather annoying that we install these executables in prefix/bin but they must be called via a symlink named 'lyx'. Then we could split the rpm package into lyx-common, lyx-qt, lyx-xforms etc. and allow to install lyx-qt and lyx-xforms in parallel. A dummy package lyx could then contain a symlink lyx > preferred frontend executable. Georg
