For some reason I used to think that Linux was destined to be a hot topic among embedded control developers, so I have been disappointed last summer, at the MASTERs (the 11th), when a couple of enthusiastic colleagues decided to try and offer a short class on Linux tools for Microchip products. The class was great, I learned a lot in that single hour and I came out with a great desire to learn more about developing code in Linux, but the attendance (statistics) turned out pretty discouraging.
If you removed the (numerous) Microchip employees, who sneaked in the room, from the total present, I counted just 9 attendees. That was out of a total pool just short of a thousand embedded control designers registered for the main event, or less than 1%!
As you probably are aware, Microchip currently does not offer any support for its software and hardware tools under the Linux operating system (although it is possible to run most of the software tools under a Windows simulator in Linux). So the little material available and presented during the class had been contributed spontaneously by a few gifted and passionate developers in the open source community. Microchip rationale, somewhat confirmed by the class attendance, is that it does not make sense to make the effort to support an operating system for a potential incremental audience of just 1% …
But you see, this is a catch-22, a vicious circle. The lack of tools attracts no users, hence no attendance to the class which, in turn, guarantees that no development will be funded…
All it would take to break the circle is a few good applications…
An announcement I received today has given me hope that things could be about to change. Canna Software has just released a new Linux application called Kpk2cmd that supports the PICKit 2 little programmer and in circuit debugger. This is great news as the PICKit 2 is one of the most popular and inexpensive tools ($34.99) ever produced by Microchip and it covers an ever larger portfolio of products from 8 pin to 100 pin, from the PIC10 to the PIC24 series.
I wish them good luck and tons of downloads, as for me, I am going to test it soon on the Linux (Ubuntu) machine that I use in the living room as my Digital Video Recorder (I use the MythTV application but that is another story…)
I will let you know how it works …
Interesting thread. I’ve often thought about adapting the C# Develop (or in Linux, the Mono Develop) environment, adapting the IDE for use with GNU C and Microchip tools. Better yet, wouldn’t it be great if Microchip worked with the PIC community on such a project in a vein similar to Microsoft’s open source initiative (e.g. nUnit, sandcastle, etc.). The usability issues MPLAB has (no, I’m not trashing MPLAB, it’s a good tool and the price is right), feature enhancement, etc. may be resolved quicker and allow the user to tailor the environment to suite their needs. I realize that MPLAB has IP build in, but it seems that there must be a way (obfuscated code in a Microchip supplied dll for example) that remain IP yet stream line the IDE. This would seem to reduce cost of development for Microchip.
By the way, will we see you at this years Master’s?
Catch-22 indeed. But I don’t understand the reluctance to release the tools/IDE source to the Linux community, especially the IDE. The tools are largely free already, and the open source improvements could feed right back into the Windows versions. For commercial product companies, using PICs with the “Pro C” would be a license issue, just as it is now — if you want to use it for a production pay or face potential legal complications. For home project folks and students and the like, what difference to Microchip’s commercial viability could it make to use the full compiler free of charge? I am reminded of Bill Hewlett and David Packard who encouraged engineers at Hewlett-Packard to take parts home for whatever use they could find — they knew that it created a more productive environment and everyone benefited. I can’t conceive of a scenario in which the Pro C tools or the IDE source, in and of itself, could mean the difference to Microchip’s commercial viability or failure. If there truly are no Linux users, then releasing the tools source will be ‘the sound of one hand clapping’; but if, on the other hand, there are people that will make it thrive, it will likely draw more users to the product line. Either way, it is a boon for all concerned. As an aside, I can tell you that I use many proprietary Windows-only tools — Solidworks, Altium, etc — but whenever there is a Linux alternative such as ModelSim, Xilinx and Altera tools, and the like, I always choose the Linux tools. And so does everyone I know.