[sane-devel] SANE protocol over USB as new USB scanner device class?
[Please CC me in all replies] On Sun, 2008-12-28 at 10:23 +0100, Julien BLACHE wrote: Paul Wise pabs3 at bonedaddy.net wrote: The way to do this would be to define some USB protocol and have it standardised by the USB Implementers Forum, then use that to influence the OEM scanner manufacturers to implement it on new scanners. There actually was a scanner/imaging device standard class defined for USB, and it never went anywhere. Bah. Any pointers on that? I haven't been able to find anything on the intartubes about such a class yet. I would propose that the SANE network protocol be ported to USB (so that You'd better start from scratch and design a real protocol that doesn't have all the suckiness and limitations the current SANE network protocol has. Fair enough. -- bye, pabs http://bonedaddy.net/pabs3/ -- next part -- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 197 bytes Desc: This is a digitally signed message part Url : http://lists.alioth.debian.org/pipermail/sane-devel/attachments/20081229/6ce1add4/attachment.pgp
[sane-devel] SANE protocol over USB as new USB scanner device class?
On Sun, 2008-12-28 at 15:28 -0800, David Poole dpoole at marvell.com wrote: http://www.usb.org/developers/defined_class/#BaseClass06h www.usb.org/developers/devclass_docs/usb_still_img10.pdf Thanks. When we were developing our scanners, we looked into the USB still imaging device. The spec is huge and, IIRC, parts of it require paying for (PIMA15740? ISO15740? I can't remember). Bugger.. We decided to come up with our own because it was faster, easier, there was no real standard anyway, and we could test both ends in-house. With our own protocol, we could do our own testing and not worry about testing interoperability with 'N' different stacks. Makes sense I guess. I'm not saying creating our own protocol was a perfect solution. It was the best solution we had at the time. But we did try to make it easy to reverse engineer. Sounds like you would be the kind of manufacturer to publish protocol specs, do you plan to do that, or have you done so? I imagine you'd get some kudos for that, just like Intel, AMD/ATI, VIA have in the graphics card space for releasing their documentation. I'd suggest hosting the specs on the SANE website (like x.org does for GPU docs) and announcements on LWN Slashdot at least. This precedent and pressure from HP/Canon etc might then be used to get other scanner OEMs to do similar and or employ developers to work on the SANE project, adding drivers before their scanners are released. Finally, the protocol specs could be distilled into sets of similar protocols and each set could be standardised into one specific protocol. Then the scanner OEMs could pick one of the standardised protocol for each scanner, depending on the complexity they need for that specific scanner. -- bye, pabs http://bonedaddy.net/pabs3/ -- next part -- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 197 bytes Desc: This is a digitally signed message part Url : http://lists.alioth.debian.org/pipermail/sane-devel/attachments/20081229/cf6a4524/attachment-0001.pgp
[sane-devel] SANE protocol over USB as new USB scanner device class?
Fran?ois Revol revol at free.fr wrote: Hi, There actually was a scanner/imaging device standard class defined for USB, and it never went anywhere. Well, just like webcams, until finally some big player switched to UVC, like Apple. My point. Except the iSight doesn't actually talk proper UVC. Does it ? Seeing how UVC is being bent over and fscked with various accessories by vendors (much like USB used to be/still is sometimes), I'm not sure having a standard class would make much difference in the end, once you've discovered all the vendor/model-specific quirks :P Just when I thought it was going to be simpler to write a webcam driver... They should be publically spanked. Fran?ois.
[sane-devel] SANE protocol over USB as new USB scanner device class?
Fran?ois Revol revol at free.fr wrote: Hi, My point. Except the iSight doesn't actually talk proper UVC. Does it ? Yep. Look at the uvc driver, you'll see the packet format for the iSight isn't standard UVC. Moreover the camera actually doesn't advertise itself as a UVC class device. The firmware extractor actually patches the firmware to have the iSight advertise itself as such... JB. -- Julien BLACHE http://www.jblache.org jb at jblache.org GPG KeyID 0xF5D65169
[sane-devel] SANE protocol over USB as new USB scanner device class?
Hi all, [Please CC me in all replies] I've been thinking it would be cool if the SANE project didn't have to write new scanner drivers for each and every new USB scanner that comes out. The way to do this would be to define some USB protocol and have it standardised by the USB Implementers Forum, then use that to influence the OEM scanner manufacturers to implement it on new scanners. I would propose that the SANE network protocol be ported to USB (so that it doesn't require TCP/IP - correct me if it already doesn't require TCP/IP) and proposed as a new USB scanner device class at the USB Implementers Forum. What do people think about this idea? Are there any people from the USB Implementers Forum here? Could this idea be extended to scanners using non-USB transports? Obviously for WiFi, Ethernet or other TCP/IP networks the existing SANE protocol could presumably be used. -- bye, pabs http://bonedaddy.net/pabs3/ -- next part -- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 197 bytes Desc: This is a digitally signed message part Url : http://lists.alioth.debian.org/pipermail/sane-devel/attachments/20081228/3207625a/attachment.pgp
[sane-devel] SANE protocol over USB as new USB scanner device class?
Paul Wise pabs3 at bonedaddy.net wrote: Hi, The way to do this would be to define some USB protocol and have it standardised by the USB Implementers Forum, then use that to influence the OEM scanner manufacturers to implement it on new scanners. There actually was a scanner/imaging device standard class defined for USB, and it never went anywhere. I would propose that the SANE network protocol be ported to USB (so that You'd better start from scratch and design a real protocol that doesn't have all the suckiness and limitations the current SANE network protocol has. In no particular order: this protocol is too tightly tied to the SANE API (which has a whole lot of limitations itself and will get overhauled at some point), it's synchronous, doing proper marshalling on the data exchanged is impossible, it uses control + data connections, ... Not a good thing to start from, trust me. JB. -- Julien BLACHE http://www.jblache.org jb at jblache.org GPG KeyID 0xF5D65169
[sane-devel] SANE protocol over USB as new USB scanner device class?
Hi, The way to do this would be to define some USB protocol and have it standardised by the USB Implementers Forum, then use that to influence the OEM scanner manufacturers to implement it on new scanners. There actually was a scanner/imaging device standard class defined for USB, and it never went anywhere. Well, just like webcams, until finally some big player switched to UVC, like Apple. Nothing stops you from lobbying your prefered vendor into quitting their proprietary protocol. Just list them the pros, like less time developping a windows only driver, Vista and Windows 7 support without writing another driver, support for all other OSes, and big advertisement from all FOSS fans. The problem is always the same with those vendors, they think having proprietary protocols and drivers will help them differentiate with added value, while all their drivers add is usually more megabytes of useless junk to install :-( Also, by doing so they are also accomplices of the microsoft monopoly, since they prevent other OSes from competing by not releasing the specs, and the lack of hw support is the number one argument from those forcing windows preloaded on PCs, because people want something that works. Fran?ois.
[sane-devel] SANE protocol over USB as new USB scanner device class?
Fran?ois Revol revol at free.fr wrote: Hi, There actually was a scanner/imaging device standard class defined for USB, and it never went anywhere. Well, just like webcams, until finally some big player switched to UVC, like Apple. My point. Except the iSight doesn't actually talk proper UVC. Seeing how UVC is being bent over and fscked with various accessories by vendors (much like USB used to be/still is sometimes), I'm not sure having a standard class would make much difference in the end, once you've discovered all the vendor/model-specific quirks :P JB. -- Julien BLACHE http://www.jblache.org jb at jblache.org GPG KeyID 0xF5D65169
[sane-devel] SANE protocol over USB as new USB scanner device class?
2008/12/27 Paul Wise pabs3 at bonedaddy.net: Hi all, [Please CC me in all replies] I've been thinking it would be cool if the SANE project didn't have to write new scanner drivers for each and every new USB scanner that comes out. The way to do this would be to define some USB protocol and have it standardised by the USB Implementers Forum, then use that to influence the OEM scanner manufacturers to implement it on new scanners. This protocol exists, it is called SCSI. Most high-dollar machines just use some sort of SCSI-over-USB wrapper, much like USB Mass Storage Class does. I can usually get a scan out of such a scanner in a few hours work. It is however, a fairly heavyweight protocol, and requires a bit of brains on the device side. For vendors with lower target prices, I'm afraid you're out of luck. We are, after all, talking about folks who can't manage to squeeze in a ROM to hold the firmware most of the time. allan -- The truth is an offense, but not a sin
