--- In ts-7000%40yahoogroups.com, "bburtan65" <> wrote:
>
> The bottom line is that if TS wants to become the de facto standard for embedded Linux SBCs then they need to insulate customers from the hardcore details of Linux so customers can concentrate on building their end-user products.
I'm pretty happy with my TS board. I've set it up with 2.6.29 kernel and lenny EABI rootfs that I've copied from my NSLU2. I'm pretty much insulated and can work with board as with any other computer. No need to setup complicated cross compiling environments. I could probably setup something like Qt with single "aptitude install qt4-dev-tools".
I'm a hobbyist so I'm using the board for non-commercial stuff. I didn't even try to use the default busybox or ts kernel. For me the shortest path to success was to get recent kernel and recent version of debian / openembedded running on the board and from there on rely on the broad userbase and available documentation of those communities. This option gives me the most options, support and continuity. I do think that could also apply on the commercial side. However, I don't know the client profile TS has. Which percentage the clients are content with the default stuff and which percentage use the board to break the boundaries using the latest stuff available. Might be that the majority of their customers are content with the default busybox for example and use it for simple C programs / shell scripts which read serial port etc.
>From my standpoint (as a hobbyist) TS is doing the wrong thing (maintaining their own kernel fork / busybox rootfs / debian rootfs). They could concentrate all that effort into getting TS peripheral stuff driver support into mainline and just riding the wave from there on. Rootfs is not a problem. Anyone should be fairly easily able to come up with a rootfs which suits their specific needs. I do think that the same applies also on the commercial side.