Re: LCC and blobs
md@Linux.IT (Marco d'Itri) writes:
> On Dec 11, Goswin von Brederlow <brederlo@informatik.uni-tuebingen.de> wrote:
>
>> > I do not believe that this is obvious. I understand that FSF disagrees,
>> > and considers firmwares to be just "data".
>>
>> Would you accept a patch for ppp of the form:
>>
>> char data[] = { 0x17, 0x23, 0x42, ...};
>> ...
>> *(int (*)(int))data(fd);
>>
>> After all, it is just data.
> No, because these bytes are executed on the system CPU under the control
> of the OS, so to me they look like software.
Where in the GPL does it say anything about being executed or where or
under which control? Where does the DFSG say something about that?
Apart from being ugly the above is perfectly legal and nothing speaks
against adding it, _provided_ this is the source. I have actually seen
GPL sources with such byte sequences in it for cases where the
toolchain couldn't emit the right opcodes.
But for anything more than a handfull of bytes I would doubt that it
is the actual source and that the data was generated and only for that
reason I would reject it.
>> The FSF also things the GFDL is free, nothing new there. I also highly
>> doubt even the FSF would consider a file stating "DO NOT DISTRIBUTE"
>> as being GPL compatible because it is "just data". And some of the
>> files in question in the kernel do.
> I was not talking about this kind of files (which obviously have some
> kind of licensing problem).
Good.
>> > Not so small: the drivers for most WiFi network adapters, many bluetooth
>> > adapters, all USB DSL modems, most (all?) DVB receivers, most 3D video
>> > cards are just the ones I can think about right now.
>> I don't know about WiFi or bluetooh but the usb dsl modems seem to
>> work fine with e.g. eagle-usb-* packages.
> You choose as an example a package which under the proposed
> interpretation of policy would be removed from the distribution.
> *ALL* drivers for USB modems are like this (I know: I have spent over
> two months making Debian the distribution with the best support for them).
Removed from main, but not from non-free and contrib. And that is as
it should be. The firmware code being uploaded is non-free or not?
>> And I haven't heart of any 3D video card that _requires_ extra
>> firmware to work at all. It is true that you need extras to benefit
>> from 3D but that doesn't make them non-functional. 2D support is good
> It makes 3D non functional (I'm talking about Radeon and Matrox cards),
> this looks like important enough for me.
It makes it useless for some people but not to others. I haven't used
any 3D in over a year and didn't miss it. I think there is a large
enough subset of Debian users that feel the same to make 2D only still
be usefull.
>> enough to install extra modules. It might be more work but there is no
>> requirement for tainting the main kernel image with non-free 3D
>> drivers.
> We are not discussing distribution of non-free drivers, but 100% free
> (GPL'ed or X11) drivers.
>
>> PS: I was going by the list of troublesome files posted month ago,
>> maybe that was incomplete but it wasn't overly long (comparted to
>> kernel size).
> It did not contain drivers which load firmwares from an external file.
> That list is much longer, especially if you add drivers which are not in
> the kernel tree.
Luckily I always bought the right hardware so far. Checking for linux
support and if extra patches/files are needed or not is one of my
deciding arguments when buying new stuff.
If you feel moving non-free firmware loading drivers to contrib is too
much inconvenience for Debians users by all means start to change
policy. I'm just interpreting it as I see it.
MfG
Goswin
Reply to: