Bonesteel Wrote: > I'm actually surprised there isn't regression testing in place already.If you > go back and check, that's not what I wrote. myself Wrote: > a major component of getting there would be a better scheme for > regression checkingThere is regression checking, but it's inadequate - I > surmise - because it didn't find these problems. I don't think Slim knew that 6 would be a problem but released it anyways. There is an extensive beta test program within the slim development group, and a regression checker. I think, though, that beta testers, either by virtue of who they are or by virtue of the fact that they run big machines, don't notice certain types of problems. So I don't think relying solely on the experience of beta testers who are all quite experienced and an old regression checking suite is enough. Which is why I volunteered to help with a better regression checker.
> I was told that I would need to tinker with my system or file bug > reports to reclaim previous levels of performance. I do not have the > time for this approach. Therefore I simply returned to an earlier > release.I sympathize with both sides of this problem. Clearly I am a user and sympathize with that side. Let me tell you about the other side. A software system like this works across a network, on at least 2 platforms (player and server). It can operate on up to 6x6 combinations of platforms (clients: SLiMP3, Squeezebox 1, Squeezebox 2, Squeezebox 3, Softsqueeze, Winamp. Server: Windows, Linux, Unix, Mac OSX, Linksys NSLU2, Buffalo Linkstation, maybe more). Diagnosing such problems is quite complex. Just showing up and saying "It doesn't work" without telling us which IT you have and how exactly it isn't working doesn't really get anyone anywhere. No one can help. Imagine calling a car dealer and saying "It doesn't work". They'd say "drive it down here and we'll look at it". Doesn't work well trans-atlantic. > the final testing step has been pushed out to the consumer.I'm not a Slim > employee and I don't speak for them, but I don't think that's fair. There are many different combinations of server software, server hardware, client hardware. It's pretty hard to test them all. Add wireless, and interference from the microwave oven in the next apartment to yours and no one can test an environment just like yours, because there is no environment just like yours. All the major configurations get a workout, but maybe not yours. > If Slim Devices published performance metrics for each release A good idea. > presumably pre-release testing is done on each platform.the resources needed > for this would be astronomical. Even the matrix I showed you above, 6 server configurations, 6 client configurations, would be a 36 test matrix. Add low memory, regular memory, fast processor, slow processor configurations, and you're at 144 combinations (well, a few less, because there is no memory size variation available for the NSLU2). I agree your need needs to be addressed. I'm just not sure how, since the requirement is more complex than you seem to envison. -- Michaelwagner ------------------------------------------------------------------------ Michaelwagner's Profile: http://forums.slimdevices.com/member.php?userid=428 View this thread: http://forums.slimdevices.com/showthread.php?t=17864 _______________________________________________ Discuss mailing list [email protected] http://lists.slimdevices.com/lists/listinfo/discuss
