Neuros Technology Forums

>But I am afraid 'ssh/nfs' would be a geek feature, and will not be in formal release (or even in, but not advertised).

The OSD supports telnet which is also a 'geek' feature, one that really is all but obselete. Substituting sshd for telnetd would not only give the same terminal functionality but also enable things like scp and tunnelling.  Scp -- secure copy -- makes much more sense than messing around with TFTP or even FTP.

From a developer's perspective this makes a lot more sense than Samba. The only part of Samba that's useful for this product (IMO) is abstracting NBNS/WINS support so we could find the Neuros by name from a Windoze machine.  A product like this is most useful when it offers web pages so I'd expect user file transfers to be mostly HTTP.

As it stands today this is a 'geek' product so you might as well just admit it and add the 'geek' feature set in.  There's no point in underbuilding the Linux until the product is developed, it just slows down work.

>I'll plod on with upacking the development tarball, but I'll do it on a spare system........

Well, the development environment wasn't as intimidating as it looked at first sight. I was a bit worried that it might screw up my other crossdevelopment environment but its actually quite benign.

One further thought about these sorts of TiVo like devices -- TiVo recently started shipping their "Series 3" HD unit.  We are rapidly switching over to HDTV  and while its clear that this OSD hasn't really got the horsepower for anything other than standard definition its a good advanced prototype platform for a future generation HD device.  This is going to be the only route for HD work in the future, everything commerical seems to be encumbered with a growing obsession with 'rights', essentially you end up paying more and more for equipment that does less and less.  Even the TiVo is gradually being infected -- I like my Series 1 because TiVo isn't upgrading the software any more so I don't have to worry about advertisement overlays, locked out or limited time recordings or any of the other 'enhancements' that they've made to the later units.

>The firmware is progressing, it's becoming more stable but there's still a lot to be done. It is a Gamma release after all. All you can do (aside from returning of course) is wait and see how the firmware progresses. (greyback)

I thought the whole point was to toss the firmware out into the environment and see what happens.

You (i.e Neuros) going to meet all sorts of people out there. Some are going to be enthusiastic hobbyists, people who may not be very experienced but are full of energy and interest. Some might be people who've spent a lifetime in product design and development, not to mention software project and project team management.  Many will be in-between and from just the sprinkling of posts in this thread its obvious that Joe Born, who thought our user base was not "geeky/tech savvy" (like Apple users???(!)*), is probably underestimating everyone a bit.  For my part, I'm comparing this unit with a Mikrotec board which I'm using for a completely different type of project (not a router, though, for those who know the 522). I'm not interested in the stability of the application(s) as much as the stability of the platform because working on those applications is a waste of time until the platform is stable.  So, getting down to basics, I'd like to know whether those crashes people report are actually crashes or system panics that cause a watchdog to trip -- and if there's a watchdog tripping then is it because some driver's holding the processor and so not letting the watchdog update. All this usually updates a log ni the /var/log/messages file or similar but we've got an environment that is missing all that good stuff. You can develop without it, you can develop with just a serial port and a few printfs (the "Bare Knuckles" school of software engineering), but you'd be crazy to do this in a tool rich environment like Linux.

Put another way.....Linux systems don't usually crash. If they do it should take but a couple of minutes to figure out why and do something about it.

I'll plod on with upacking the development tarball, but I'll do it on a spare system........

*Yes, I know people who use Apple systems as BSD platforms with a nice interface (i.e."geeky"). But that's not Apple's primary market.

Lemajors's comments are right on the money. My initial impressions of the user interface were:-

-- You can learn a lot from owning a TiVo (which, BTW, is a Linux box). The graphics and user interface is really well thought out on the TiVo, the OSD isn't anything like as good.
-- You have to update the video during the vertical blanking interval, if you don't you get an annoying flicker as that part of the image is changed
-- The remote control is anemic. I don't think its the remote's fault, it appears that the software in the OSD has very high latency and chokes if you have to keep the remote sending (like, for example, when you're scrolling an address entry to 255 -- ever thought of going down as well as up?). Needless to say, the TiVo's remote is responsive (and its got nicely thought out gimmcks such as backing up the fast forward when stopping to compensate for the human reaction time and system lag).

From a developer perspective I'm a bit disappointed. I'm used to working on these types of small systems and I'd expect the platform to be a lot more stable at this stage of the game -- the application might crash, things may not be implemented but the system should alway stay up. One thing that would contribute to this is a more modular approach to the software -- I've yet to poke around the sources but it appears that the application is one monolithic thing like a typical Windoze app so, not surprisingly, it behaves just like one.  (Its also as tedious to develop and debug as one......) What I'd expect to see is a bunch of functional programs doing things like recording and playing media, each with their own configuration files and assorted support libraries, plus applications to edit those configuraton files with the whole lot driven from a gui that shells these programs as needed (note that a webserver is an example of this type of thing). I'd also expect to see a lot more use of scripting.

Other things I've noticed are that I can't automount my Compact Flash card and I can't alter fstab to uncomment or change the lines to try doing so -- the file may be marked read/write but the filesystem appears to be read-only.  So it looks as if I have to build and burn a new image to try anything -- but the image is burnt into on-board Flash, something that's OK for occasional use but dicey if you're doing it all the time (much better to park your development filesystem on a removable Flash drive like the CF). One example of the problems of updating is from going from .20 to .23 my CF moved from hdc to hda....there's no rhyme or reason to this, it just says to me that the build environment is suspect.

For the future, note that ARM seems to do a nice line in brochures and technical glitz but I've always found that their processors (ARM5, ARM7) suck compared to MIPS, PowerPC or even x86 running at the same clock rate.  Their main claim to fame is small footprint on the silicon and, as they say, you get what you pay for.......