[BRLTTY] make drivers a library?

[Samuel Thibault]

Note btw that I'm not saying nvda would only this, I'm just suggesting
having the alternative to pass through brlapi or not.

Hermann, le Tue 22 Sep 2009 15:35:31 +0200, a écrit :
> > (yes, I believe we got to a very simple way to achieve that, but two
> > applications is still too much for casual users).
> 
> The problem here, I think, is, that Windows users are not familiar with
> Brltty, and its concept to run as a service, allowing to use several
> applications with it.

Yes, even SAPI acts as a library and thus is not so complicated even if
voices have to be installed separately.

> They want to stick to the concept,

I don't think they necessarily primarily want to stick to that (even if
there's something like that), we have to admit that it's easier to just
install one software and not have to manage starting one then the other,
especially since there are interferences with jaws and one may have to
stop jaws and start brltty+nvda, then stop brltty+nvda and restart jaws.
It's not handy to not have nvda coupled with brltty.

> > As a result, I yet more and more believe that brltty should export its
> > drivers through a library too, even with an simple interface similar
> > to libbrlapi/libbraille.  Else I fear dispersion of hardware support
> > efforts.
>
> What would be the difference? In my understanding, one has only to
> copy/install a driver for only one display, looking like "libalva.so" or
> "libht.so" (ht = Handytech). Am I right?

In the principle, yes, but in practice the current driver model includes
only the code for the particular device, and there is a bunch of
additional helper functions shared between all of them (serial, usb,
bluetooth support etc.) which would need to be packaged separately too.

> > What do people think?
> 
> If this makes the use of Brltty easier for Windows users, we should
> think about it, since I don't believe that too many companies will develope
> special drivers for NVDA.

I believe it can be useful in other cases too. We could for instance
think about a braille backend in the Debian installer which would not
need the burden of the brlapi protocol. That could also be used for
grub's braille support etc.

Samuel