It appears that commit d4ae5fb0b5d1ae4270b3343509e8bd2d529aa291 (here: https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=commit;h=d4ae5fb0b5d1ae4270b3343509e8bd2d529aa291 ) breaks the '-l:' command line option of ld when given an absolute file name. Since -l:/absolute/path/to/lib.so is not a documented feature (ony -l:lib.so is documented), I was wondering if this issue could be considered as a bug, or was done on purpose?
It was a deliberate bug fix. *** This bug has been marked as a duplicate of bug 17532 ***
I understand bug 17532 is a deliberate fix. However, I was talking more precisely about using an absolute path name. Traditionaly, absolute paths are considered as such and do not require any "search path". I noticed that using an explicit -L/ in front of a -l:/absolute/path.so make it work, which I find a bit weird. For instance, if there happen to be a /foo/absolute/path.so, using -L/foo -l:/absolute/path.so would not give the expected results. I think that either -l:/absolute/path.so should raise an error, or be treated as such. What do you think?
I don't see -l:/... being much different to -l/... For example, -L. -l/c will load ./lib/c.so or ./lib/c.a if such files exist. I don't think ld should warn in either case.
*** Bug 17800 has been marked as a duplicate of this bug. ***
*** Bug 18172 has been marked as a duplicate of this bug. ***