Re: netscape status?
email@example.com (Amy Fong) writes:
> Greg Stark <firstname.lastname@example.org> wrote:
> >So a while back there was a discussion of why netscape core dumps so
> >frequently, and the suggestion was that there was some kind of egcs
> >optimization bug in the X libraries. Has this been dealt with? Because I still
> >see netscape crashing every few minutes and it's becoming increasingly
> I hate to bring Redhat into this but their netscape package actually
> uses a wrap-around script to work around some of the libc6 2.1 bugs. I have
> been using a modified version of it to work with my potato system and it's
> been behaving pretty well so far.
Debian's wrapper script does a lot more than just this. It also sets
MALLOC_CHECK_ to ignore netsacpe's habit of double-freeing pointers for
example. It also supports having default configurations in /etc/netscape4 and
it starts a new window if there's an existing netscape running.
But neither the libBrokenLocale nor the MALLOC_CHECK_ setting seems to be
enough. Netscape crashes for me 9 times out of 10 the instant I hit a page
with java, it also crashes or freezes randomly at other times. And it
occasionally just starts chewing up all of memory until my machine crashes
unless I kill it fast enough.
So now I'm at a loss, if the xlib patches didn't solve the problem the only
solution I can see is downloading the supported libc5 package. Geez, I
remember doing actual java development using the libc5 netscape on an 8M
machine, these days I can barely run netscape in 16M on the same machine,
libc6 is just bigger I guess. Having both libc5 and libc6 binaries running on
the same machine isn't going to be better though.
If we can't work out the libc6 netscape problems it would be nice if we could
get libc5 netscape packages. But I guess I shouldn't whine too much, my own
packages are languishing until the next break in my work.