This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
About decimal floating point support
- From: Wu Zhou <woodzltc at cn dot ibm dot com>
- To: drow at false dot org
- Cc: gdb-patches at sourceware dot org
- Date: Tue, 13 Jun 2006 23:51:23 +0800 (CST)
- Subject: About decimal floating point support
Hi Daniel,
We ever talked about adding decimal floating point into GDB. And you
provided a lot insightful suggestion for my first patch. In my post at
2005-10-18:
http://sources.redhat.com/ml/gdb-patches/2005-10/msg00148.html
http://sources.redhat.com/ml/gdb-patches/2005-10/msg00149.html
I revised the patch and the testcase. I believe it can address most of
your concern.
But there is one concern not addressed: you ever proposed to move
libdecnumber to the top level of the repository, and use that to implement
DFP support in GDB. At that time (Oct, 2005), DFP support for gcc is not
matured yet. The libdecnumber code may change itself. So I didn't bother
to adopt you method at that time and put my focus on resolving other
problem you pointed out. Now since DFP is already in the 4.2 development
tree of gcc. So I guess it is time to revist this.
What is your current thought about this? Do you still think it is more
feasible to put libdecnumber at the top level repository? I am now
considering copy the code of libdecnumber (which is at the top level of
gcc repository) to the top level of gdb repository. and use that to
re-implment DFP support. If the code of libdecnumber is changed, we can
synchronize that to gdb repository. Then gdb developers don't need to
maintain these sensitive numeric code.
What do you think on this? Your comments are highly appreciated!
Regards
- Wu Zhou