[linux-dvb] Multiproto API/Driver Update

Markus Rechberger mrechberger at gmail.com
Mon Sep 8 14:21:21 CEST 2008


On Mon, Sep 8, 2008 at 1:14 PM, Simon Kenyon <simon at koala.ie> wrote:
> Markus Rechberger wrote:
>>
>> On Sun, Sep 7, 2008 at 9:15 PM, Simon Kenyon <simon at koala.ie> wrote:
>>
>>>
>>> Steven Toth wrote:
>>>
>>>>
>>>> A big difference between can and will, the em28xx fiasco tells us this.
>>>>
>>>>
>>>
>>> just wondering if your tree will go any way towards resolving that
>>> little problem?
>>>
>>
>> I don't see any fiasco nor problem here, it's more comfortable to have
>> the em28xx in an extra
>> tree and do ongoing development with it. People who followed the
>> development  during the last
>> few years know that newer things got added and newer chips are
>> supported by it, it will continue
>> to evolve anyway.
>>
>> The driverwork is currently around 30% of it, patching applications
>> and providing full support from
>> the driver till the endapplications is what everything's focussing on
>> mcentral.de not just driver only
>> development.
>>
>> We do dedicated application support for several customers too (analog
>> tv, radio, dvb integration in their
>> business applications).
>> There are many new products coming up, adding support for them is on
>> the roadmap.
>>
>> Markus
>>
>>
>
> i have had the device for a while now
> i originally got it because the wiki showed that support was being worked on
> then that all changed and you moved your stuff elsewhere
>
> i really don't understand your rationale. i really don't.
> i'm not trying to start a flame war, but your arguments for having your own
> tree and doing stuff user side makes no sense
>
> let my try to explain:
>
> you say:
>
>> it is more comfortable to have the em28xx in an extra tree
>
> it may very well be comfortable for you. but it is impossible for me
> i have lots of different devices. i can choose between using your tree and
> having support for one of them.
> or i change choose the linuxtv.org tree and have supoort for all the other.
> how is that "more comfortable"?
> i really don't understand that argument
>
> you then say:
>
>> newer things get added and newer chips are supported by it
>
> how does that help me? it doesn't, as far as i can see
>
> next you say:
>
>> the driverwork is currently around 30% of it, patching applications and
>> providing full support from the driver till the...
>
> apart from the little speedbump that is multiproto vs. s2api most apps don't
> need "supporting". they are already supported by their respective
> communities.
>
>> we do dedicated application support for several customers...
>
> now i think we get to the heart of the matter.
> you have built up a nice little business providing support for the various
> devices. you have customers who (in my humble opinion misguidedly) think
> that by having support in your tree that they can proudly boast "linux
> supported". all i can say is well done. congratulations on building a
> sucessful business.
>
> but please do not mistake that for providing linux support for the em28xx
> devices in linux. what you are providing and "linux support of the em28xx
> class of devices" are not one and the same thing. for better or worse, linux
> support means that at some point (and it may well be one year from now) it
> will be in the kernel.
> have you not seen all the software that has come and gone because linus will
> not put it in the tree. right now the only route to doing that is to get it
> in the linuxtv.org tree. from there it stands some chance of making it into
> the kernel. i seriously doubt that any other approach has any chance of
> success.
>
> this may not be of concern or interest to you. well so be it. but it is
> important to me.
>

the code on mcentral.de didn't write itself, it was mostly written
from our side.
If you're going to pay my monthly insurance for working together with
linuxtv.org (something has to
compensate the slowdown of the development due those endless wanted
discussions) I'll be glad to
go that way.

Application development and driver development is right now at an
acceptable speed, noone bothers
about fixed drivers there which improve the signal strength etc.

The code is already used with BSD [1] now and will be used by our OSX
driver in near future.

Markus

[1] http://netbsd-soc.sourceforge.net/projects/dvb/



More information about the linux-dvb mailing list