Re: Speech in the kernel

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]


On 15 Mar, Dave Mielke wrote:
> On Mon, 15 Mar 1999, T.Pospisek's MailLists wrote:
>>I actually can't see any strong arguments against besides kernel bloat
> The difficulty of making quick enhancements.

I don't know any software with quicker release cycles than the kernel.
Try to beat that if you can.

And even if, it doesn't matter. You can keep your speech-patch
development completely separated from kernel development. All it'd take
the user to have a speech is "patch -p0 < speech_patch" and there he

You're not dependent on kernel release schedules at all.

> The lack of portability to other platforms and operating systems.

Are you certain? I would guess that pretty much all systems need to
have a auditory way of communicating an error, in case the video is
defect. So it must be one of the first output systems to work (before
the video system actually).

>>and the problem of having to recognize speech hardware as early as
>>possible in the startup phase 
> This, of course, is an argument for, and not against, putting support into the
> kernel.

No this is against. Because you need to have an initialised PCI or even
worse an ISA bus to use most soundcards. And if you have to wait for
that (read in many cases isapnp) then you have filesystem access which
is initialised as the last system, and that means you can as well use a
userspace solution which will be loaded from the harddisk as the first
app by init.

> Another one would be the ability to react immediately when characters
> are written to the screen, rather than just passively analyzing /dev/vcsa0, by
> scanning its entire content, every so often.

This can be done in userspace. You just have to have a hook from the

>>- Use the internal speaker for output: 
> I don't think its possible to hook up headphones to it on most platforms. This
> would be an issue if a person is in an open work area. Also, on most platforms
> it is not possible to adjust its volume.

Maybe you are missunderstanding me. I mean the "beep, beep"
internal speaker, that comes with every PC. Nothing to do with

>>the internal speaker should be
>>  really easy to use, since it is an absolute standart on all PCs. 
> Yes, but that standard does not extend to workstations of any kind.

Are you sure? See above.

> Also, even
> on PCs, the way in which the standard is applied is different from one
> operating system to another.

That doesn't matter.

>>  don't know about lilo, but maybe it could be even used by lilo (!!!)
> LILO can be told to beep the internal speaker when it issues the prompt, and to
> write its output to a serial port, but it cannot, without extensive
> modification, be told to send its output to the internal speaker in the form of
> speech. Just little old LILO, with no kernel support whatsoever, is all that's
> running in the CPU at that point.

Yes, but it can beep. And that's the point.

>>- There's an existing kernel patch that uses the internal speaker for
>>  sound production.
> I don't know, and would be interested in knowing more about, this patch.

I don't have the URL handy, but it should be easy to find. You can
maybe have a look at or in the kernel patches area.


             Tomas Pospisek - Freelance: Linuxing, Networking
             - Internet Services in Graubuenden/Switzerland

[Index of Archives]     [Linux for the Blind]     [Fedora]     [Kernel List]     [Red Hat Install]     [Red Hat Watch List]     [Red Hat Development]     [Gimp]     [Yosemite News]     [Big List of Linux Books]