Problems compiling Shibboleth 2.4.3

[Cantor, Scott]

On 8/29/12 5:40 PM, "Law, Bob" <Robert.Law at wolterskluwer.com> wrote:

>I¹m trying to compile Shibboleth on Solaris 10.

You really should be using the latest version, and that goes triple if you
literally can't get the old one to build. That's the obvious next step
unfortunatey.

>  I have already compiled the following libraries
>Log4cpp-1.1rc1

Unneeded.

>Httpd-2.0.64

You do know this is basically all but EOL at this point? They never
released a major security fix for 2.0 (2.0.65 was never released) and it's
basically dead. Apache is not a package you want to be on a dead version
of.

>Mod_perl-2.0.3
>Tomcat-connectors-1.2.35

Neither has any relavance to the SP build.

> 
>My make file for Shibboleth looks like this:
>    cd shibboleth-2.4.3; XERCESCROOT=$(XERCESCROOT); ./configure \

You won't need XERCESCROOT anymore.

>Here are the errors that I am getting.  One question I have is if
>mod_perl even needs to be in apache.  We used it in prior verisions of
>Shibboelth because of our WAYF and a redirect.  Since the discovery
>service is in java running on tomcat,
> can¹t I also have the ³protected pages² be jsp pages on tomcat and
>perform my redirect there?

You don't "protect" discovery pages, by definition. They have to be
public/unauthenticated.

> 
>mod_apache.cpp: In member function 'virtual const char*
>ShibTargetApache::getScheme() const':
>mod_apache.cpp:365:66: error: 'ap_http_method' was not declared in this
>scope

Don't know what that is, though it rings a bell. I would probably start
cross-checking all the makefiles and logs and make sure it's pulling in
the right Apache headers. Then I'd go find the function and figure out
where it should be declared and then what's not getting included that
should be.

>In file included from mod_shib_20.cpp:68:0:
>mod_apache.cpp: At global scope:
>mod_apache.cpp:1637:5: error: expected primary-expression before '.' token

Those I might guess at if I knew what compiler was involved, which you
didn't mention. If it's a fairly recent gcc, that would be a compiler bug
I ran into that requires working around and I can describe some possible
ways.

-- Scott