addition to the bugdb form

[Dean Gaudet <dg...@arctic.org>]

The form, when asking what version the user is using, should ask: "Did you
compile this yourself?  Where did you get the source/binaries?"

I've just peeked at the debian packaged apache and they #if 0d some
critical code ... which will result in ErrorDocument problems like the
ones introduced in something like 1.2b6.  It'd suck if we have to deal
with error reports on this sort of thing. 

I'm trying to find the linux package which advocates kill -9. 

Dean

Re: addition to the bugdb form

[Marc Slemko <ma...@worldgate.com>]

On Tue, 26 Aug 1997, Dean Gaudet wrote:

> 
> 
> On Tue, 26 Aug 1997, Marc Slemko wrote:
> 
> > Let's alias /doc to /usr/doc.
> 
> Omygawd.
> 
> > Let's hardcode in -lcrypt in the htpasswd makefile for fun.
> 
> This is needed by any linux system running glibc (which I believe debian
> does).  We fixed this in 1.3.

But if it is needed then add it to the libs, don't hardcode the makefile.

> 
> > Let's not give any feedback to the Apache people, we can just fix bugs here
> > and feel special.
> 
> Did they really fix bugs?  I didn't look hard enough i guess.

Not too much, but a bit.

> 
> > Let's restart Apache after log rotation by using killall -HUP.
> 
> Omygawd.
> 
> > Let's encourage people to make htpasswd setuid root.
> 
> Omygawd.
> 
> > Let's make a /usr/include/apache (!??!)
> 
> To build dynamic modules from somewhere else maybe?

/usr/include/apache is a symlink to somewhere like /usr/src/apache.  Sigh.

> 
> > Let's have some conf file for some silly program to scan
> > ftp.apache.org:/pub/dist for new copies of Apache.
> 
> Ugh.

I should have noted, in case it isn't obvious, that /pub/dist doesn't
exist.

> 
> > Now I remember why I don't use Linux... I don't mind the OS and a
> > new kernel can be quite refreshing.  It is using it that is the
> > problem.
> 
> Yeah this is why I stick with redhat, afaik it has fewer of these problems
> if you use redhat supplied rpms only.  I got a response from them already
> they're fixing the kill -9 problem.  They asked if there's a way to know
> when apache has exited ... unfortunately we don't do anything useful.  do
> we remove the pid file?  My log-locking stuff would achieve this ... 

We should remove the PID file.  Hmm... better idea... write -1 to it.
Yup, I'm aware of what that would do when scripts used it.  <g>

Seriously, removing the PID file may be smart, but it would break some
things that expect it to be there.  OTOH, they are already broken just
less.  The lockfile stuff would work but is a bit more complex for use in
shell scripts.

Re: addition to the bugdb form

[Dean Gaudet <dg...@arctic.org>]

On Tue, 26 Aug 1997, Marc Slemko wrote:

> Let's alias /doc to /usr/doc.

Omygawd.

> Let's hardcode in -lcrypt in the htpasswd makefile for fun.

This is needed by any linux system running glibc (which I believe debian
does).  We fixed this in 1.3.

> Let's not give any feedback to the Apache people, we can just fix bugs here
> and feel special.

Did they really fix bugs?  I didn't look hard enough i guess.

> Let's restart Apache after log rotation by using killall -HUP.

Omygawd.

> Let's encourage people to make htpasswd setuid root.

Omygawd.

> Let's make a /usr/include/apache (!??!)

To build dynamic modules from somewhere else maybe?

> Let's have some conf file for some silly program to scan
> ftp.apache.org:/pub/dist for new copies of Apache.

Ugh.

> Now I remember why I don't use Linux... I don't mind the OS and a
> new kernel can be quite refreshing.  It is using it that is the
> problem.

Yeah this is why I stick with redhat, afaik it has fewer of these problems
if you use redhat supplied rpms only.  I got a response from them already
they're fixing the kill -9 problem.  They asked if there's a way to know
when apache has exited ... unfortunately we don't do anything useful.  do
we remove the pid file?  My log-locking stuff would achieve this ... 

Dean

Re: addition to the bugdb form

[Marc Slemko <ma...@worldgate.com>]

On Tue, 26 Aug 1997, Dean Gaudet wrote:

> Already mailed them.  And I mailed the debian people about their #if 0
> stuff. 

Oh.  My.  Gosh.  Time for a good rant.

ftp://ftp.debian.org/debian/unstable/source/web/apache_1.2.4-1.diff.gz

Hey, we don't understand this code lets attempt to hack it and comment it
out.

Let's put our config files in /etc/apache.

Let's have a /usr/lib/apache.

Let's include mod_usertrack to make everyone happy with lots of cookies.
May not end up being run though with their funky dynamic system.

Let's include perl_module for fun.  Nothing against mod_perl, simply
that when someone says they are running "Apache" without knowing
they did anything else...

What is:

  int i = atoi("1")

supposed to mean?

Let's put CGIs in /usr/lib/cgi-bin.

Let's uncomment BindAddress * to make people even more confused.

Let's have /var/lib/apache for the proxy cache.

Let's alias /doc to /usr/doc.

Let's hardcode in -lcrypt in the htpasswd makefile for fun.

Let's change the name of the binary to apache from httpd and make all
sorts of other silly changes like httpd_monitor to apache_monitor, even
though it doesn't even work on our platform since we are using shared
memory.  But hey, since we are just changing httpd to apache, we will
still call it httpd_monitor, just have everywhere refer to it as
apache_monitor.

Let's not give any feedback to the Apache people, we can just fix bugs here
and feel special.

Let's restart Apache after log rotation by using killall -HUP.

Let's complain about having /etc/apache as a bogus directory even 
though we say to use it.

Let's encourage people to make htpasswd setuid root.

Let's put mime.types in /etc.

Let's make a /usr/include/apache (!??!)

Let's have some conf file for some silly program to scan
ftp.apache.org:/pub/dist for new copies of Apache.

Let's update /etc/passwd by creating a new one then moving it over,
ignoring the concept of file locking.

Some of the above are more important than others.

Now I remember why I don't use Linux... I don't mind the OS and a
new kernel can be quite refreshing.  It is using it that is the
problem.

Re: addition to the bugdb form

[Dean Gaudet <dg...@arctic.org>]

Already mailed them.  And I mailed the debian people about their #if 0
stuff. 

Dean

On Tue, 26 Aug 1997, Alexei Kosut wrote:

> On Tue, 26 Aug 1997, Dean Gaudet wrote:
> 
> > On Tue, 26 Aug 1997, Dean Gaudet wrote:
> > 
> > > I'm trying to find the linux package which advocates kill -9. 
> > 
> > Damnit.  It's redhat, the most common linux distribution, and it's not
> > just apache that they kill -9.  Here is one of the standard functions that
> > most of the /etc/rc.d/init.d scripts use:
> 
> Cute. And really annoying. Are you going to email the redhat people?
> 
> -- Alexei Kosut <ak...@organic.com>
> 
> 

Re: addition to the bugdb form

[Alexei Kosut <ak...@organic.com>]

On Tue, 26 Aug 1997, Dean Gaudet wrote:

> On Tue, 26 Aug 1997, Dean Gaudet wrote:
> 
> > I'm trying to find the linux package which advocates kill -9. 
> 
> Damnit.  It's redhat, the most common linux distribution, and it's not
> just apache that they kill -9.  Here is one of the standard functions that
> most of the /etc/rc.d/init.d scripts use:

Cute. And really annoying. Are you going to email the redhat people?

-- Alexei Kosut <ak...@organic.com>

Re: addition to the bugdb form

[Dean Gaudet <dg...@arctic.org>]

On Tue, 26 Aug 1997, Dean Gaudet wrote:

> I'm trying to find the linux package which advocates kill -9. 

Damnit.  It's redhat, the most common linux distribution, and it's not
just apache that they kill -9.  Here is one of the standard functions that
most of the /etc/rc.d/init.d scripts use:

# A function to stop a program.
killproc() {
        # Test syntax.
        if [ $# = 0 ]; then
                echo "Usage: killproc {program}"
                return 1
        fi

        # Save basename.
        base=`basename $1`

        # Find pid.
        pid=`pidofproc $base`

        # Kill it.
        if [ "$pid" != "" ] ; then
                echo -n "$base "
                kill -9 $pid
        fi

        # Remove pid file if any.
        rm -f /var/run/$base.pid
}

kill -9 baby.  I can't believe it. 

Dean