Paul wrote: > Hi All, > > We are working for the infrastructure setup for desktop testing. Now I'd > like to collect the feedback about remote testing requirement for > desktop testing. You know, the testing community in OpenSolaris > community has provided some hardwares and one program called "Test > Farm", then it can let community user leverage Sun's hardware for > Solaris kernel/driver development and testing. For desktop testing, > because all of our applications are GUI, the requirement is a little > different from that. So I'd like to get your feedback if we need to do > something like this way. If I can get enough positive feedback about it, > I can discuss with team about how to setup it. > > The questions can be like this: > 1. Do you think it is necessary to provide community user hardwares for > desktop application remote testing? > 2. If yes, could you please list some reasons? > > I'd like to set time limit from Aug. 26 to Sep. 1. Please give me your > feedback. > > Thanks, > Paul. > _______________________________________________ > desktop-discuss mailing list > desktop-discuss at opensolaris.org > 1.) Yes, there's probably a good handful of people that need to check their own projects against the current GNOME tree on Solaris.
2.) I think it's necessary to provide access to machines that have a point release intended for external usage before it goes live so that developers can make sure their applications are ready even if they're not on their own systems. This would imply providing VNC access to test themes against multiple versions of JDS as well as changes to underlying GNOME dependencies that would result in visual complications for example. How one goes about allowing multiple concurrent sessions is up to those responsible for providing the hardware, which leaves room for two options. VirtualBox instances that automatically rollback at a specified time or shell accounts with the ability for a userland VNC server process to be used. I don't see there being any downsides for limited per-server accounts on bare metal. VirtualBox has its overhead and performance differences, though generally on modern hardware shouldn't be an issue. On the other hand, I think shell access on both instances would be necessary if someone wants to throw some generic GTK+ code around for a spin. Just tell me what you think. James
