[Microkernel-devroom] [Talk-Proposal] Rathaxes - A DSL for device driver development, why and how?
Hello, Our turn to check in! First things first: - title of your talk: A DSL for device driver development, why and how? - your full name: Rathaxes; - duration of your talk: 30 minutes, without questions. Our abstract: Device drivers are notoriously known to be a critical and prominent part of modern operating systems. We will draw up a summary of the current state of the art in device driver development: tools, methodologies and approaches which try to mitigate the problem. Then, we will present our take on it: a domain specific language with its own aspect-oriented programming paradigm, and its compiler, Rathaxes. We will show you interesting details of an implementation, in the Rathaxes DSL, of the e1000 ethernet driver for Linux. We will conclude with how Rathaxes fits into the current state of the art, our vision, and next steps. And that's it! If you have any question, please let me know. We are also hanging out on freenode in the #rathaxes channel. Our homepage is: http://www.rathaxes.org/. Best -- Louis Opter
Hello Louis, though I'm personally interested in the topic of your work, could you please elaborate on how it is connected to microkernel and component-based OSes? Regards -- Christian Helmuth Genode Labs http://www.genode-labs.com/ · http://genode.org/ https://twitter.com/GenodeLabs · /ˈdʒiː.nəʊd/ Genode Labs GmbH · Amtsgericht Dresden · HRB 28424 · Sitz Dresden Geschäftsführer: Dr.-Ing. Norman Feske, Christian Helmuth
On Fri, 7 Dec 2012 17:11:56 +0100 Christian Helmuth <christian.helmuth@genode-labs.com> wrote:
Hello Louis,
though I'm personally interested in the topic of your work, could you please elaborate on how it is connected to microkernel and component-based OSes?
Hello Christian, Right, our project isn't specific to microkernels and component-based OSes, but it does apply to it, entirely. Coming across the list of devrooms, we just figured out that this devroom was the best fit. AFAIK, there is no other room that much focused on the “lower-level” parts of operating systems? We also discussed a little bit with Martin Děcký last year and it seemed that we raised some interest, so here we are! Best -- Louis Opter
Hello folks,
We also discussed a little bit with Martin Děcký last year and it seemed that we raised some interest, so here we are!
Yes, I can confirm that. I was and I am still interested in the topic. Even if microkernel systems can certainly provide more isolation and protection to/against the device drivers than monolithic systems, any means of making the drivers correct-by-design and simplifying their actual development is helpful. Louis, I just suggest that you focus your talk a little bit more on the (potential) benefits of a device driver DSL specifically for the microkernel systems. Frankly, stressing in your abstract that a prominent case study of yours is based on Linux is just a bit provocative in the context of a microkernel devroom :) On the other hand, Christian, if we are not in a serious time pressure already, we might give the topic a chance (and a slot). I can even vouch for them in the sense that the HelenOS team is indeed interested in adopting a technology like this in the future -- as we strongly believe that implementing compatibility wrappers for Linux/*BSD/whatever device drivers is not the best thing to do. However, we do not chair this year's devroom, thus of course the final call is yours. M.D.
On Sat, 08 Dec 2012 00:16:54 +0100 Martin Decky <martin@decky.cz> wrote:
Hello folks,
We also discussed a little bit with Martin Děcký last year and it seemed that we raised some interest, so here we are!
Yes, I can confirm that. I was and I am still interested in the topic. Even if microkernel systems can certainly provide more isolation and protection to/against the device drivers than monolithic systems, any means of making the drivers correct-by-design and simplifying their actual development is helpful.
That and also, make driver code more re-usable and more easy to collaborate on; but I'm starting to spoil you :>.
Louis, I just suggest that you focus your talk a little bit more on the (potential) benefits of a device driver DSL specifically for the microkernel systems. Frankly, stressing in your abstract that a prominent case study of yours is based on Linux is just a bit provocative in the context of a microkernel devroom :)
Actually, I think there was a misunderstanding on our side while reading the devroom's CFP. While the room is driven by micro-kernel development teams, we thought that the list of topics would not be limited to micro-kernels. We'd be happy to focus our talk more on micro-kernels, but we aren't there yet, our language isn't even entirely defined at this point. (We are a small, but stubborn, team side project). We want to target any kind of kernel, and our first generated code appears to be for Linux because it was simpler and more common for a lot of people. It was definitely not a provocation attempt :)
On the other hand, Christian, if we are not in a serious time pressure already, we might give the topic a chance (and a slot). I can even vouch for them in the sense that the HelenOS team is indeed interested in adopting a technology like this in the future -- as we strongly believe that implementing compatibility wrappers for Linux/*BSD/whatever device drivers is not the best thing to do.
Thanks, now I understand why we might be a little bit outside the expected scope of the devroom, but I still think this is the best place to do our talk! Best regards -- Louis Opter
Hi, thanks for describing the incentive behind your proposal. I definitely did not want to rule out your talk - I was just curious why you chose this devroom. Greets -- Christian Helmuth Genode Labs http://www.genode-labs.com/ · http://genode.org/ https://twitter.com/GenodeLabs · /ˈdʒiː.nəʊd/ Genode Labs GmbH · Amtsgericht Dresden · HRB 28424 · Sitz Dresden Geschäftsführer: Dr.-Ing. Norman Feske, Christian Helmuth
participants (3)
-
Christian Helmuth -
Louis Opter -
Martin Decky