Re: [linux-dvb-maintainer] Re: [PATCH] b2c2: make front-ends selectable and include noob option

[Prakash Punnoor <prakash@punnoor.de>]

[-- Attachment #1: Type: text/plain, Size: 3039 bytes --]

Am Mittwoch Dezember 7 2005 08:56 schrieb Patrick Boettcher:
> Hi,
>
> On Wed, 7 Dec 2005, Johannes Stezenbach wrote:
> > I think b2c2-flexcop-pci uses a 240K dma buffer, whether you
> > save a few K in demodulator code doesn't mean much.
> > The saved memory will be similarly unnoticable to the user as
> > if you would go and scatter #ifdefs all over tuner-simple.c.
> >
> > But I'm neither the author nor the maintainer of the b2c2-flexcop
> > driver, you better ask Patrick if he likes it.
>
> There will be at least two new devices in the future, which again will
> need (at least) two new demod(and maybe tuner)-modules.
>
> Prakash, if it is just the i2c_xfer failed, that could be easily turned
> into a debug-message, but I rather have these as errors, because they are.

Well aren't these expected errors? You want to attach an non-existent 
front-end and the driver finds out it isn't there? IMHO it doesn't look very 
professional if some strange errors are scattered in the log. I think if you 
don't find *any* front-end there should be perhaps an error or note to the 
user, but not because half a dozen of front-ends drivers can't find the hw. 
At least please make the errors verbose and not cryptical, if you really want 
them.

If these errors are not expected, they have been here since months and 
probably should have been fixed. But I think I remember someone saying that 
they are expected once I reported them...

> I don't see the need for the ifdef - on the contrary: the number of people
> asking for which demod they shall load dropped significantly after
> FE_REFACTORING starting one year ago - now reintroducing that again is not
> a good idea, IMO.

Well, I got twice annoyed by missing reverse dependecies introduced by new 
front-ends. Once the fw loader and once the new lgt330x(?) front-end. If new 
front-ends wouldn't go in automatically as rev dependency this wouldn't break 
for me (and everyone alse using b2c2 in general and not only the new 
frontends), which is another reason, if you don't care about the bit of code 
saving.

I really don't see the big problem of the ifdefs as the are in the *header* 
files as it is just checking whether something is in or not - as the user 
selected thus emulating that a probe for a front-end failed... If the .c file 
is coded properly there is no chance of breaking - otherwise you'd have the 
same problem with the unpatched source (leaving the module problem out for 
now). (Yes, I know ifdefs in the code as such are bad, but I noted that in 
conjunction with the fw loader, which was the only "dirty" ifdef.)

And as I said I added an *option* for unsure users to include all supported 
front-ends, which is another one of your arguments. Selecting it by default 
(I don't know how to do this) would be what you want then. Wouldn't this make 
all happy (if Michael additionally fixes the module stuff)?

bye,

-- 
(°=                 =°)
//\ Prakash Punnoor /\\
V_/                 \_V

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]