Polling for the Future… TCP/IP stack

Now that eMOS for AVR is in beta, time to polish off the crystal balls some more. I have a consultant looking at CANLIB for AVR already. I’d think that a robust, fast, small, full-feature TCP/IP stack is high on the list of things we should do. Problem there is there are 4 qualifiers there, and as they usually say, you can get 2 out of 3, or 3 out of 4. So what features would you want in a TCP/IP stack?

A compiler tie-in is that someone suggested that we provide some sort of API so that a program can take a web page HTML with embedded references to C variables, e.g.

…// html stuff …
…. @foo:bar …

where bar is a local variable in foo, and the API would expand the values at runtime. I am not sure what kind of variable reference we can allow yet, but you get the idea. Yay or nay?

Comments and suggestions always welcome. Contact me at richard at imagecraft.com

This entry was posted in new product and tagged , . Bookmark the permalink.

2 Responses to Polling for the Future… TCP/IP stack

  1. What I would want is some something that would help you to interface the TCP/IP stack to the hardware. The AVR doesn’t come with an ethernet interface, so everybody is probably going to have something different. This would, it seems to me, be the biggest support headache.

    Also, what do you have in mind to make me to want to purchase this product. I am currently using Ethernut…(that I purchased the hardware for here originally). The TCP/IP stack comes with Ethernut, for free. And it seems to offer just about every service you could want for embedded Internet.

    I don’t mind paying for software, but if I can get something for free, I will go that route. And, despite it being free, Ethernut does have fairly good support. And it is compatable with ICC…

  2. imagecraft says:

    re: James
    Yes, fair question. Someone also asks basically, what is our value proposition if there are cheap solution in the form of the Microchip chipset which can be used as an accessory from even an AVR.

    First: the low level ethernet interface. Yes, of course this part will be made easily changeable. The Realtek chip probably covers >50% of the needs. Then if we add support for the Wiznet and a couple others, it should be sufficient for most if not everybody.

    Second: we believe we can make a stack that really takes advantages of our deep understanding with how C works, and what our customers need – hence these postings. We know there are free alternatives out there in the form of uIP, Ethernut, Microchip etc. but software engineering is all about making the right trade off, and the ones that others made may not necessarily be the ones that our customers really want. Also, we believe that ultimately, a total package solution would be beneficial to our customers.

Leave a Reply

Your email address will not be published. Required fields are marked *