[Bug ld/10518] [regression] handling of wildcards in version scripts

[ryanryan52 at gmail dot com]

------- Additional Comments From ryanryan52 at gmail dot com  2009-08-20 10:17 
-------
(In reply to comment #1)
> The linker script shown has two global wildcard matches.  As the ld info doc
> says, this is slightly crazy.  They both match "test_new", so which one should
> ld choose?  The most specific match?  How do you do that in general?  eg. is
> "tes*" a better match than "*new" for "test_new"?  (If we make a special case
> for "*", why stop there?)

the reason to make the special case is because the docs say that the special
case is made.
from the ld info page:

You can bind all otherwise unspecified symbols to a given version node by using
`global: *;' somewhere in the version script.

that is exactly what this example is doing, however ld is not behaving in the
way it is documented (afaict). and there are real life things (wxwidgets, for
example) that currently depend on this (documented) behavior.

the differences between test* and *new are really irrelevant here, the problem
is not related to that, it's related to the use of "global: *", and ld not
acting on that as documented.


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=10518

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.