[3487] in Athena Bugs
2.1.2 beta2 vax-bsd - initial comments
daemon@ATHENA.MIT.EDU (Ken Raeburn)
Tue Oct 24 13:00:36 1989
Date: Tue, 24 Oct 89 13:00:11 -0400
From: Ken Raeburn <Raeburn@ATHENA.MIT.EDU>
To: saber-bugs@ATHENA.MIT.EDU
Cc: bugs@ATHENA.MIT.EDU
The installation went fairly smoothly.
There was a problem with install_commands checking for writable
directories; the test was failing when I tried with NFS directories.
Commenting out the check was sufficient.
I still have the same complaints about the startup messages as before:
It prints out your customer support number. If the customer support
number and original email address were configurable (i.e., not checked
against the password), we could simply the number here at Athena,
since we want to filter the bug reports through our own bug-tracking
system, and don't necessarily want people to go calling you when it's
a local problem or something we know how to fix. (We'd also want to
omit the "Become a Registered User" bit, I think.)
My first gripes/bug reports about this version:
1. I'm glad to see that saber is no longer faulting on gcc symbol
tables; do you have plans to support gcc symbol tables soon? (I
personally use gcc almost exclusively when I have the option,
so....)
2. I have an object file (okay, it's a g++ object file, with a static
object that requires a constructor, so the symbol table is a bit
funny, though I didn't use `-g') which saber immediately unloads:
1 -> load foo.o
Loading: foo.o
Unloading: foo.o
2 ->
G++ uses a new symtab type (which is recognized by the loader) to
designate constructors and destructors. (Actually, it's turned into
an accessible vector at link time, and the name signifies whether it's
a constructor or destructor or what; but this is beside the point.)
Saber unloads the file without explaining why.
3. As you've noted in the release notes, `struct foo const *' doesn't
work. I tend to use this a lot, especially dealing with X toolkit
programming; do you expect to have this one fixed soon?
4. When I ran into #3 in nxsaber, I fixed the file and clicked on
`load' again. I don't think I inserted anything with the mouse,
but I got:
2 -> load ) {
Cannot open '/afs/athena.mit.edu/user/r/raeburn/)'.
Cannot open '/afs/athena.mit.edu/user/r/raeburn/{'.
3 ->
I had this happen a couple of times, with different random garbage.
5. It might be helpful for nxsaber to give you a display someplace of
what the current directory is (either for the program or for
loading source files, I don't know; does saber revert directories
when popping back to top-level?).
6. Under OSF's `mwm' window manager at least, nxsaber doesn't come up
with a tall enough window. The top of the button for `delete' in
the `debugging' box is just visible.
I'll put out announcements ASAP, try to get you some comments on the
new version.
-- Ken
