[Bug 435829] Review Request: tomcat6 - Apache Servlet/JSP Engine, RI for Servlet 2.5/JSP 2.1 API

[bugzilla at redhat.com]

Please do not reply directly to this email. All additional
comments should be made in the comments box of this bug report.

Summary: Review Request: tomcat6 -  Apache Servlet/JSP Engine, RI for Servlet 2.5/JSP 2.1 API


https://bugzilla.redhat.com/show_bug.cgi?id=435829





------- Additional Comments From dwalluck at redhat.com  2008-04-02 12:09 EST -------
Jason Corley suggested in comment #20 doing something with the unversioned jsp
and servlet provides. Are these backwards compatible or not? Otherwise, why even
keep them at all. I am not sure that we can version the other provides since the
version is the tomcat version, which would not be the case with some other API
provider, presumably.

The license was ASL 2.0 in the past rpms. I changed it back to what JPackage
had. This seems correct according to the Fedora wiki. I think rpmlint is just
wrong here.

I will fix the coreutils BuildRequires.

I don't think we can mark the doc webapp as doc. We can't mark any content
needed to run as doc. Docs must be 100% optional.

We usually check for the existence of /usr/share/java-utils/java-functions
first. I can do this, but why? We already Requires: jpackage-utils, so it must
exist.

/srv is fixed

The other two are more complicated. I find java using java-functions, but I left
most of the script the same as JPackage. It is probably fine, but it could use
java-functions more.

About the initscript, I changed most of what I could. The reason we can't
necessarily use the built-in killproc function is that: (1) we had to call
"stop" to stop the server---but Jason Brittain told me that tomcat should do the
right thing when it receives a KILL signal (I haven't verified this) (2) if the
PID file is missing, it will try to find the tomcat6 process (clearly wrong,
it's some java process), and I don't know how the PID file will go missing in
practice.

For startup, maybe we could use the daemon function? Right now it's hardcoded to
call runuser if it exists and su if not.

The main issue is that the initscript is not using the standard functions making
the code large and difficult to maintain. For example, the initscript output was
broken because it was overriding some built-in functions from
/etc/init.d/functions. Jason Corley told me that most of this was due to
incompatibilities in the Fedora and RHEL functions, but I would rather write the
script in the standard way and deal with this problem if/when it arises for us.


-- 
Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.