RE: [certi-dev] CERTI billard test / http_proxy

[Gotthard, Petr]

Hi Martin,

> > (1) Please check also status of the http proxy variables:
> > CERTI_HTTP_PROXY and http_proxy
> 
> Et voila, tu as fait mouche ! I was unaware that 'billard' not only
> honours the $CERTI_HTTP_PROXY but also queries $http_proxy. The latter
> was set, the former not, therefore I felt totally innocent  :-)
> Is it really useful to honour the 'universal' $http_proxy if there's a
> special $CERTI_HTTP_PROXY for exclusive use with CERTI ?

We understood $http_proxy as a global variable that should affect all
applications, including CERTI. The $CERTI_HTTP_PROXY is meant to change
the global behavior, i.e. enable proxy if $http_proxy is not set, or
disable proxy if $http_proxy is set.
https://savannah.nongnu.org/patch/?func=detailitem&item_id=6561#comment1

I tried to describe this behavior at [currently not accessible]
http://www.nongnu.org/certi/certi_doc/User/html/certi_HTTP_proxy.html

It's meant to be more convenient for inexperienced users that don't run
own RTIG as they don't have to set yet another proxy variable.


I will try to enhance the error reporting so that is easier to discover
the CERTI proxy misconfiguration.

> I probably won't be able to get my hands onto the Linux machine before
> tomorrow afternoon (we all live almost in the same timezone, don't
> we ?  ;-)

I guess so, although the certi-devel e-mails sometimes come also at 1AM.
;-)


Best Regards,
Petr