Limits to what can be done without source

[Andy Green]

Les Mikesell wrote:
> On Wed, 2006-05-17 at 14:16, Andy Green wrote:
> 
>> ABI churn is not the only problem with binary blobs.  Point in case I 
>> saw on this list in the last couple of days, Adobe Acrobat blew chunks 
>> on a double free.  This is not an ABI problem but a hidden bug in the 
>> binary blob.
> 
> It's not hidden to everyone so that's not a reasonable description
> of the problem.  It's just an ordinary bug to be fixed by the
> responsible party.  Give them some reason to care about fixing
> it - like a platform that has a reputation for cooperating with
> other suppliers and a large user base - and the responsible
> parties will take care of their parts.

Except that users of binary blobs are often in for long waits to get a 
new version with their bug fixed.  Often the Linux version of their 
offerings can be a bit of a redhaired stepchild, eg, Flash and Skype. 
That double free would have had a patch out fixing it that evening had 
it been FOSS.

>>> Provide a documented and unchanging interface so if something works
>>> today it will still work next week.
>> That does not follow for the same reason... a stable ABI would be nice 
>> but that's not what one can expect with Linux.  It won't guarantee 
>> binary blobs becoming paragons of coding virtue and to provide immortal 
>> functionality either.
> 
> There are broken binary blobs and there are binary blobs that
> are perfectly fine.  It doesn't make a lot of sense to
> overgeneralize about them.  There's a lot of crud available
> in source too.

The "overgeneralization" is because your point seemed to be that ABI 
breakage was the major or only problem faced by binary blobs.

> It's the issue that keeps them from being fixed.  I don't see
> similar problems happening with OS X for example.  There are
> normal bugs that show up, but once fixed they don't reappear
> on every release.

Hum OSX runs on a tiny landscape of platforms, all documented and 
understood by the OS vendor.  Before the release, they can consider to 
make a formal effort to test for regressions.  It's telling you 
something that you don't have the same problems on OSX+Apple HW, but 
it's hard to see what the lesson really is for the sprawling, open-ended 
platform, BIOS, peripheral mixture supported on Linux.

What I have had -- am having right now typing this -- a good experience 
with is the Xorg nv driver.  For me it works very well and has always 
done so for the functionality it claims to support.  It seems a good bet 
to me that if nVidia donated the sources it currently keeps private into 
that project then things will go on in the same smooth way but with 
complete support.  Not least because Fedora would ship with it all in 
the box.

-Andy
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 4492 bytes
Desc: S/MIME Cryptographic Signature
URL: <http://listman.redhat.com/archives/fedora-list/attachments/20060517/4738fa71/attachment-0001.bin>