Skip to main content ↑ ↓
Topic: Web Platform v2? (Read 214315 times) previous topic - next topic

Re: Web Platform v2?

Reply #46 –
Is all the work really needed?
http://www.plugcomputer.org/
buy one and see what improvements can be made, but for a web server connected on the public network, there really is a need to be able to handle the network load  by the use of multi threaded systems, I really do not see the 'pic' as a logical choice.
If something is going to be built, then it needs to be useful and a talked about product.

Re: Web Platform v2?

Reply #47 –
> If something is going to be built, then it needs to be useful and a talked about product.

No - it needs to sell, and the Web Platform is already a good seller (approaching 300 sold to date); so someone must find it useful whether or not it's talked about :-)

Re: Web Platform v2?

Reply #48 –
300 pcs is not a big number for the amount of work that went into it. (even if there was a $25us margin on it.)
 Also you should consider  'Netiquette'.
Slapping underpowered and potentially insecure equipment on the net does not do anyone any favours, then there are the issues related to IPv6.
By standardizing on a more powerful processor that runs an off the shelf OS you  free yourself to do what the project was intended to do , and that is experiment, link items to it, write web code etc, but more importantly it frees the 'maintainers' of the hardware up to do more profitable things, which is where the REAL money and sales are, such as attachments.(not to mention the enjoyment)

 if you want a 'special' function/ IO, etc then you write a driver that drops into the kernel, which can be driven directly from a multitude of programming languages, that ANYONE not just c programmers can use.
 

Re: Web Platform v2?

Reply #49 –
[quote author="hardcore"]
 Also you should consider  'Netiquette'.
Slapping underpowered and potentially insecure equipment on the net does not do anyone any favours, then there are the issues related to IPv6. [/quote]

It depends on where you use it.  On my LAN it is useful for me and being behind my firewall I do not think it endangers anyone else.  (I think IPv6 is still a long way off -- since good NAT has been developed I think the push for IPv6 has lost momentum.  It will come eventually but I think it will be a long time -- if ever -- before IPv4 only equipment will no longer work on the net.)

[quote author="hardcore"]
By standardizing on a more powerful processor that runs an off the shelf OS you  free yourself to do what the project was intended to do , and that is experiment, link items to it, write web code etc,... [/quote]

Again, it depends.  I do not think I would have a use for one of those.  I do however like the idea of adding some Ethernet/IP connectivity to some of my dsPIC projects and so have a use for the current version -- which is why I bought two.

[quote author="hardcore"]
300 pcs is not a big number for the amount of work that went into it. (even if there was a $25us margin on it.)...
... but more importantly it frees the 'maintainers' of the hardware up to do more profitable things, which is where the REAL money and sales are, such as attachments.(not to mention the enjoyment)[/quote]

I guess that depends on what the motivation/enjoyment of the project is -- money?  I didn't think it was but then I wasn't involved.

[quote author="hardcore"]
 if you want a 'special' function/ IO, etc then you write a driver that drops into the kernel, which can be driven directly from a multitude of programming languages, that ANYONE not just c programmers can use.[/quote]
 
Yes, if you don't want to program in C or assembly then you are probably going to need a more powerful platform.  For most of what I want to do or experiment with though this would be overkill. 

When I have to configure/fix/rebuild my wireless router I am working on a platform closer to what you are describing.  It runs a nearly full fledged distribution of Linux (OpenWRT) -- it actually has a more powerful CPU than I think the first computer I ran Linux on had.  It is a great network platform.  I doubt we could build a 400MHz MIPS, 64MB RAM, 32 MB FLASH, five gigabit port VLAN capable Ethernet switch, 802.11b/g/n wireless, USB host, etc., etc. system plus case and power supply for $69.99 shipped.  If you want to do web design with high level scripting languages this is a great platform to start with.  It even has a couple slide witches, a few push buttons, and about half a dozen LEDS for a user interface.  However it does not have nearly the collection of low level peripherals that the dsPIC does like ADC, DAC, PWM, SPI, I2C, CAN, analog comparators, large number of digital IO, etc., etc. that make it great for hardware projects.  This is what sets the microcontrollers apart from the 32-bit embedded Linux systems -- hardware peripheral support.  They are two very different platforms -- different tools for different jobs.

I find the discussion of this new M3 based system interesting but I don't think I would end up getting one.  That doesn't mean other people wouldn't end up getting one and find it useful.  It does seem like more of a saturated market though.

[quote author="hardcore"]
Is all the work really needed?   http://www.plugcomputer.org/
buy one and see what improvements can be made, but for a web server connected on the public network, there really is a need to be able to handle the network load  by the use of multi threaded systems, I really do not see the 'pic' as a logical choice.[/quote]

I think you're right about this.  It would probably be hard for us to beat one of those for this type of platform.  If this is what you're after just get one of these.

[quote author="hardcore"]
If something is going to be built, then it needs to be useful...[/quote]

Agreed.  But what useful means and to who is the question.

I think we need to define what the goal of the project (or the goal of the next version of this project) is otherwise feature creep and good ideas will pull it down a path that was not intended.  But that's just my opinion, take it for what its worth.


-Eric

Re: Web Platform v2?

Reply #50 –
Let's face the fact that the web platform is neither a mega seller nor runs any killer app as of today.
For a few $$ more you will get full blown embedded Linux platforms (i.e. WRT-54GL or NGW100).
If you thrown in an extra US$ 55 you'll get a Plugcomputer and for about US$ 130 extra you will get
a Beagleboard (new version with Ethernet) with 3D graphics and high-quality audio support. There
are many other options for off the shelf low-cost Linux platforms ...

Like hardcore said, if you are looking at an application ready Linux supported platform then the web
platform is definitely not the right choice ... for now.

On the other side isn't DP all about (re)making/hacking electronic thingies and gadgets more or less from
scratch - regardless if they are remakes or new ideas - with the purpose of learning how they work,
how to do it (better/simpler) and sharing the knowledge and experience with anyone who is interested
and having fun (and for DP to make money - at least enough to finance more projects)?

Sure, designing a platform for Linux or Windows Embedded CE/Mobile is one option but it's not a good
starting point if you want to learn about how to implement TCP/IP and a (simple) real time OS and how
they work. uIP and/or lwIP and FreeRTOS for example are much better starting points as they come
with hardly any ballast/overhead, simple examples and are well documented.

When looking into small solar powered (self sustained) appliances with wireless networking capability
Linux/Windows CE based platforms may be no choice at all.

On the other side performance and price will be no issue when chosing one of the many (low-end)
32-bit Ethernet MCUs released over the last couple of years (based on ARM Cortex-M3, AVR32, MIPS M4k,
Coldfire cores).

Web Platform v2 could become a well selling base board for a wide range of wired and/or wireless network
applications even if no Linux/Windows CE support is available and if the price will staybelow US$ 50 (single qty)
and following features will be made available:
- flexible and plenty of I/O (USB 2.0, UARTs, SPI, I²C, PIO, EPMP ...) on headers convenient to stack expansion/breakout boards
- 10/100BaseT interface (with the option to disable the power hungry PHY - regardless if on-MCU or external)
- footprint for WiFi module (i.e. MRF24WB0Mx)
- microSD adapter and (optional) footprint for serial Flash/SRAM memory
- flexible power supply options (USB, external, LiPol with solar charger) - PoE could be an option
- a solid and well structured (easy to understand and expand) sample (RTOS based network) application - starting point/frame for application development
- free tool chain with good peripheral library support
- good basic documentation and a HowTo implement applications on the web platform 

P.S. I wrote this post while/before Eric made his last post ... to me the essential point seems that the original idea of the web platform is maintained and carried on without boggling the project down by overloading it with features and making it too complex ...

Re: Web Platform v2?

Reply #51 –
As it is now the current webplatform looks to me like a small "swissknife" for DYI, students, enthusiasts and even engineers.
It is small but it is quite flexible and not power hungry, add to that all the goals described here by IPenguin and i believe that the WP V2 can be serious leap forward.

If you want to look at this project from a lucrative point of view, than in my opinion the extra development effort should be made by the entity that seeks a nice revenue. Consider that this entity will have to pay some extra coders to:  improve/implement  the HW capabilities of the system (BTW Ian payed a coder to re-write the TCP/IP stack just to get it GNU),
                                     use an OS shell to make it more appealing to different skill level programmers
                                     develop the "all in one template" to interface the OS and the platform
                                     create and maintaining a team of HW & SW support programmers/ developers to keep it all from getting to "fat" and remain flexible and not power hungry

                                     ... enforce IP security ... implement properly buses ... study particular scenarios ...

                                     and of course document it :)

To make it short: lots of money and resources  to be spent and will it be still GNU?

The concept of WP 2 discussed here has a well defined direction and it has reached this point because of constructive criticism/input.
From my point of view all the prerequisites for this project are here already (M3 and AVR32 look like THE candidates) and from what it look like until now WP2 is going to be a very nice  swissknife.

Re: Web Platform v2?

Reply #52 –
Quote
BTW Ian payed a coder to re-write the TCP/IP stack just to get it GNU

We ported the uIP GNU stack to the web platform (kind of, a volunteer actually got it going and improved it), but it's not fair to say we rewrote it. The vast majority of the work was done by hobbyists.

The truth is, the web platform v1, like all DP projects, are designed for me :) I have an itch, a project I want to do or something, so I scratch it.  Making it available to other people is a huge bonus, and maybe someone else wants to hack around with it too.

Someone mentioned up thread one of my concerns: there is a lot more stuff in the bigger dev-board space, and it's going to be much harder for us to compete on price.
Got a question? Please ask in the forum for the fastest answers.

Re: Web Platform v2?

Reply #53 –
Two me there are two quite different ideas being here
1) Full Ethernet Connected Embedded Computer
- basically off the shelf hardware; whether a modified router or a plug computer; gumstix beagleboard etc
- complex hardware; cheap if using off the shelf consumer electronics; expensive if you want to build your own custom hardware

2) Embedded micro-controller with Ethernet connectivity
-hardware: micro-controller eval boards or Web Platform V1
-simple hardware which can be built at home
-BOM cost much lower especially in small volume

Personally I would just like to see a more advanced web platform board; low parts count; lots of fun population options.  I personally see PoE opening up a lot of applications for this thing.  When you don't need a wall wart for power it makes it easier to put these things outside or in remote areas (still need Ethernet of course).  Another nice part of the micro-controller approach is that its easy to have wide temperature operation which is handy for remote sensing.

Scott

Re: Web Platform v2?

Reply #54 –
Hmm I have several web platforms, of the original edition and I have a new BeagleBoard-xm.  The work to integrate either into a project is very different.  The BeagleBoard is a bit pricey to control a toaster, and getting the whole software chain working and up to date is a lot of work.    I can enable quite a few simple projects with the web platform for the cost of a BB-xm.  Also the amount of work required to get it working is significantly less for the web platform.  On the other hand I have a project that I am aiming for with the bb-xm, which is well beyond what a PIC could do.  Requires multiple connections and serious processing power.  So I got the BB-xm which is a lot of fun and very capable.   But for hitching up a pico weather station or other simple device control, and monitoring device with a simple website, the web Platform is a lot easier to do.  Yet a full blown system has to be maintained to safe gaurded because it is open to attack, and has the ability to be easily hacked.   The web platform, is trivial to take down with a DOS attack, but behind my private firewall, it is just fine, and mapped to a random port on the external net it works just fine for me.  Yes anyone can overloaded, but if I do my integration well enough who would ever care to try, and how would they know they are doing it?  (I am not going put an unprotected page out on port 80, nor am I going to response to every port sniffer out there which seems to be about 1/3 of the traffic to my router.)

Re: Web Platform v2?

Reply #55 –
I would love to see something that would run Linux with 100bt ethernet for under $50

Re: Web Platform v2?

Reply #56 –
Quote
I would love to see something that would run Linux with 100bt ethernet for under $50

Check out gumsticks (gumstix?), they're kind of like that. Of maybe a hackable router.

The web platform with simple, inexpensive chips is $40, I don't we'll be able to design a multi-layer board for a BGA Linux ARM and 100MBPS ethernet for $10 more.
Got a question? Please ask in the forum for the fastest answers.

Re: Web Platform v2?

Reply #57 –
[quote author="ian"]
Check out gumsticks (gumstix?), they're kind of like that. Of maybe a hackable router.
[/quote]

Gumstix are cool. If you have sources, you can even develop some expansion boards for them. Other options are Beagleboard, Hawkboard... there are lots of these around, starting proce is ove 100$ though.

Re: Web Platform v2?

Reply #58 –
i did not want to run a large distro or any thing like that maybe a simple arm 7 and an external 8MB sram boot from SDcard

Re: Web Platform v2?

Reply #59 –
I'm really interested in that project, and I've done some poking around. IT indeed would be very cool, and a great learning tool. Here's what I've found, any corrections and extra input is welcome, I could totally be wrong on all these points.

As I understand it, Linux requires a MMU, and ARM7/9/cortex 3 (?) doesn't have that. You have to get to the really big chips, and last time I looked they were all BGA package. BGA requires a multi-layer board and fancy soldering.

8MB (bytes..) of any ram is going to be a ton of address lines, and if it's DDR it requires a multi-layer board (from every estimation I've heard).

The 100MBPS etherenet is no problem, lots of even low-end ARM have that.

uC linux maybe different, but probably in ways that moot the coolness of the project.
Got a question? Please ask in the forum for the fastest answers.