[cgl_discussion] RE: MV open projects analysis

Steven Dake sdake at mvista.com
Tue Jun 24 14:54:31 PDT 2003


Patrick Mochel wrote:

>Hi there.
>
>  
>
>>PDF is updated too at:
>>http://www.osdl.org/docs/cgl_20_p1_project_list___pdf_format.pdf.
>>    
>>
>
>A couple of points on the "Hot Device Identity" Requirement, from a 2.5 
>developer standpoint.
>
>udev, and associated sysfs utilities, are intended to become the de facto
>standard for supporting persistant device data, including naming. There is
>a lot of work going on in this area, though more is greatly welcome (hint, 
>hint ;).
>
>It does currently only work for a subset of devices, but that is not a
>long-term limitation. As more device subsystems are converted to the 2.5 
>driver model, more subsystems will be supported by udev. Some patches are 
>pending (e.g. input devices), and more are yet come.
>
>As for MontaVista's SDEF utility, I strongly suggest you reconsider it's
>entry in the spreadsheet. First of all, the intent of the work is
>completely unaligned with the short-term and long-term goals of the
>relevant kernel developers, and is not supported at all by said parties.  
>While it may suffice as a solution for 2.4-based distribution, it will be
>made irrelevant by udev and friends in the 2.6 time frame. 
>  
>
The intent of the work is entirely aligned with the 2.6 kernel.  
MontaVista is using standard interfaces, such as sysfs and 
/sbin/hotplug.  The intent of system device enumeration framework is to 
provide a complete solution on 2.5 kernel+ without any major 
modifications to the kernel.  I really don't see any problem with 
competing projects such as Nokia's HDI project or Greg KH's udev project 
being listed in the PoC references.  Competition is _what_ makes Linux 
better.  What project you choose to use is certainly your decision, 
however, there are specific customer requirements that are not met by 
the current solutions available which is why MontaVista is actively 
investing in this area.

>Secondly, it's vaporware. It has not been released, and there is no
>guarantee that it will be. It is shamelessly self-promoting, and in poor
>taste, IMO. No where else in the document are entries for ficticious
>projects. I suggest removing it for the sake of consistency and honesty.
>
>  
>
It is true that no software has yet been published.  When MontaVista 
does publish, it will be under GPL/LGPL (or BSD, the decision isn't 
final) license and be an excellent solution for system device 
enumeration.  During the PoC process (a previous meeting), we had 
discussed MontaVista's solution for inclusion in the PoC references.  As 
I recall, we decided not to include the link in the PoC references until 
published source code is available.  I'll bring this up in tomorrow's 
conference call.

Thanks for pointing this out.
-steve


>	-pat
>
>
>
>  
>




More information about the cgl_discussion mailing list