From time to time, one ruby cgi scripts gives me the message:
[BUG] unknown node type 0
ruby 1.8.0 (2003-03-03) [i686-linux]
Observations:
strikes really randomly: the system run perfectly for days, then this
message comes
seems to strike only when there is a “here document” (x = <<- EOF) somewhere
in the script.
goes away if I add / remove empty lines around the “here document”, or in
fact do some completely unrelated formatting stuff.
never shows up on my production machine (good but only on my test
machine. May be related to different compiler / libraries:
gcc 2.95.3 + libc-2.2.5 : no problem
gcc 3.2.2 + libc-2.3.1 : on the machine which has that bug.
The two systems have exactly the same Ruby 1.8.0 preview 2, compiled from
source without any fancy option (./configure ; make ; make install).
Is there anybody having the same problem so that we can share observations and
eventually make a sensible bug report?
I had the same problem with several Ruby versions (1.6.7, 1.6.8,
1.8.0-pre1, 1.8.0-cvs). Now I have 1.8.0-pre2 and so far was unable to
reproduce it. When I was experiencing the bug, adding empty lines worked
for me, too. I sent the program causing the bug to appear to Guy, but he
was unable to reproduce it
Ruby was compiled using various versions of gcc 3.x (starting with 3.1
IIRC) and glibs 2.2/2.3.
The good thing is, I notice that the bug appears more and more rarely
with each Ruby release.
···
On Wed, 9 Apr 2003 14:07:54 +0900, Francois GORET wrote:
Is there anybody having the same problem so that we can share observations and
eventually make a sensible bug report?
From time to time, one ruby cgi scripts gives me the message:
[BUG] unknown node type 0
ruby 1.8.0 (2003-03-03) [i686-linux]
There are some reports like this from gcc 3.x users. I’m not sure
this is caused by gcc 3.x bug or ruby 1.8 bug. No clue yet.
I had problems on Solaris building 1.8.0 snapshots with less that
3.x. I succeeded with 3.2.2. Do you have values for x in “3.x”
that are known to fail? Have you any test cases for this bug so I
can try to reproduce it on the machines for which I have built a 1.8
snapshot?
matz.
Hugh
···
On Wed, 9 Apr 2003, Yukihiro Matsumoto wrote:
In message “[BUG] unknown node type 0” > on 03/04/09, Francois GORET fg@siamecommerce.com writes:
In message “Re: [BUG] unknown node type 0” on 03/04/09, Hugh Sasse Staff Elec Eng hgs@dmu.ac.uk writes:
I had problems on Solaris building 1.8.0 snapshots with less that
3.x. I succeeded with 3.2.2. Do you have values for x in “3.x”
that are known to fail? Have you any test cases for this bug so I
can try to reproduce it on the machines for which I have built a 1.8
snapshot?
I’m using gcc 3.2.3 without any problem. But I can’t tell whether
3.2.3 is OK, or just lucky I am. Any report is welcome.
I had problems on Solaris building 1.8.0 snapshots with less that
3.x. I succeeded with 3.2.2. Do you have values for x in “3.x”
that are known to fail? Have you any test cases for this bug so I
can try to reproduce it on the machines for which I have built a 1.8
snapshot?
I’m using gcc 3.2.3 without any problem. But I can’t tell whether
3.2.3 is OK, or just lucky I am. Any report is welcome.
Just to clarify: Is that really 3.2.3? http://gcc.gnu.org/index.html
would suggest this is not out yet?
(Meanwhile I’m trying to rebuild rubicon on an alpha to get some
more reports out to you, on how that performs.)
matz.
Hugh
···
On Wed, 9 Apr 2003, Yukihiro Matsumoto wrote:
In message “Re: [BUG] unknown node type 0” > on 03/04/09, Hugh Sasse Staff Elec Eng hgs@dmu.ac.uk writes:
I believe he’s using the following
batsman@kodos:~$ apt-cache show gcc-3.2
Package: gcc-3.2
Priority: standard
Section: devel
Installed-Size: 5504
Maintainer: Debian GCC maintainers debian-gcc@lists.debian.org
Architecture: i386
Source: gcc-3.2 (1:3.2.3ds6-0pre7)
Version: 1:3.2.3-0pre7
···
On Wed, Apr 09, 2003 at 08:06:35PM +0900, Hugh Sasse Staff Elec Eng wrote:
On Wed, 9 Apr 2003, Yukihiro Matsumoto wrote:
I’m using gcc 3.2.3 without any problem. But I can’t tell whether
3.2.3 is OK, or just lucky I am. Any report is welcome.
Just to clarify: Is that really 3.2.3? GCC, the GNU Compiler Collection - GNU Project
would suggest this is not out yet?
(Meanwhile I’m trying to rebuild rubicon on an alpha to get some
more reports out to you, on how that performs.)
===========
Debian sometimes packages things directly from the CVS.
but only on my test
