This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: glibc-2.2.5 bug triggered by Java interpreter
- From: Tom Tromey <tromey at redhat dot com>
- To: "Ulrich Weigand" <Ulrich dot Weigand at de dot ibm dot com>
- Cc: java at gcc dot gnu dot org, gcc at gcc dot gnu dot org
- Date: 02 Oct 2002 14:25:43 -0600
- Subject: Re: glibc-2.2.5 bug triggered by Java interpreter
- References: <OF9AF5D7D8.919DA8C1-ONC1256C46.0051650B@de.ibm.com>
- Reply-to: tromey at redhat dot com
>>>>> "Ulrich" == Ulrich Weigand <Ulrich.Weigand@de.ibm.com> writes:
Ulrich> While this bug is already fixed in the current glibc CVS head,
Ulrich> the question remains why gij is doing that; seeing as '$' in
Ulrich> the library name has some special semantics for dlopen even
Ulrich> with that bug fixed (it can trigger some keyword replacement
Ulrich> AFAIK), couldn't this have unexpected side effect if a class
Ulrich> has just the right name to trigger one of those keyword
Ulrich> checks?
I wasn't aware that this could happen. It doesn't seem to be
documented on my RHL 7.3 box :-(
Anyway, it seems reasonable to me for us to chop off the class name at
the `$' before passing it to dlopen. This gives a little less
flexibility in naming the .so, but I don't think this loss is very
important. Any comments on this idea?
Meanwhile, please submit a PR for this. We don't want to forget about
it.
Tom