[Vimoutliner] OT: Wi-Fi & me [Was: Re: The easy way to add new VO file extensions]

[Tim Roberts]

Matej Cepl wrote:
> 2) Microsoft is riding straight into the same mess we are in, except much
> worse -- switch from 32-bit to 64-bit for Linux is just recompilation and
> fixing bugs (there is a lot of them -- I know it pretty well, because
> there are so many people in Red Hat who are fixing them). Boring stuff but
> easily doable in reasonable time.
>    For Microsoft, on the other hand, it is pure hell -- apparently THEY
> don't have source code for most drivers, so when people will ask
> for support of some obsolete hardware with 64-bit Windows (and they
> will -- after all when you are 90% monopolist, you should support 100%
> of all hardware in production), they have to either persuade manufacturer
> to create a new 64-bit driver for some piece of sh*t which is long out of
> production (if the manufacturer is still in the bussines), or they have
> to go the same way we did -- signal analysis and dissasembling binary code.
> For hundreds of drivers at once! Good luck.
>   

Although this is off-topic for this list, your comments are quite
correct.  The problem is compounded by some of the decisions Microsoft
has made.  For example, Microsoft's 64-bit compilers do not support
inline assembler.  At all.  x86 assembly code wouldn't have worked
anyway, but by not supporting it at all, they're forcing people to
refactor their drivers to put the assembler code into separate source
modules.

Further, you cannot install a driver on a Win64 system unless the driver
package is digitally signed with a Verisign commercial certificate. 
Self-generated certificates are not good enough.  Only corporations can
get such certificates, and they cost about $395 per year.  This is
proving to be a serious burden for many independent driver writers.

We'll see how this all plays out.

-- 
Tim Roberts, timr at probo.com
Providenza & Boekelheide, Inc.