[ome-devel] Launching VisBio from the OME web interface

[Curtis Rueden]

Curtis Rueden wrote:

> I'll get to the support for session keys next week.


Ok, I added support in VisBio for passing a session key instead of 
username as a command line parameter. To support this in ome-java, I 
added a setSessionKey() method to org.openmicroscopy.ds.RemoteCaller and 
org.openmicroscopy.ds.XmlRpcCaller. To take advantage of it in the web 
interface, I modified OME/Web/VisBio.pm to use $session->SessionKey() to 
grab the active session, and pass that instead of a username.

I tested the above changes and they work well (images just pop up in 
VisBio now, sans user interaction), so I committed them to CVS. Let me 
know if anyone has any problems with it. I did notice that VisBio does 
not seem to download the Sample.ome image correctly for some reason (but 
all other images I've tried work), so I'll be looking into that.

Chris Allan wrote:

>On Linux we have the handy "hostname -f" however that doesn't exist on
>BSD systems, of which OS X's userland inherits so much from. As I recall
>there's a Perl module called Sys::FQDN or Sys::Hostname::FQDN, can't
>remember which, that does the lookup properly (ie. against the inet_*
>structures of a bound socket). I'll look at this a little closer later
>in the week.
>  
>

I was under the impression that we are planning to add an entry for the 
OMEDS FQDN in the database, in which case it would be unnecessary to 
require the Sys::Hostname::FQDN Perl module. But if that is not the 
case, I can try it out to see if it generates the FQDN properly on my 
installation.

Thanks to Chris's advice about /etc/hosts, hostname -f now produces the 
FQDN, but Sys::Hostname's hostname() subroutine still does not. I read 
up on Sys::Hostname, and it seems that it does not obtain a FQDN, just 
the hostname like "hostname" with no arguments does. So for now, I coded 
VisBio.pm to use `hostname -f` first, and only use Sys::Hostname's 
hostname() call if the first method fails.

-Curtis