Re: [orca-list] Punctuation, capital letters, exchange of characters and strings, generally error in the design of Orca

[Michael Whapples <mwhapples aim com>]

On Wed, 2008-04-09 at 16:00 +0200, Milan Zamazal wrote:
> > > > > > "MW" == Michael Whapples <mwhapples aim com> writes:
>
>     MW> I feel something such as speech-dispatcher is sufficiently core,
>     MW> and may be used by minimal apps (eg. brltty, yasr, etc) that to
>     MW> some (and in some cases be difficult) to have a python
>     MW> interpretor available. This I feel is a good reason for trying
>     MW> to avoid using python for something such as speech dispatcher.
>
> I can imagine adding Python interpreter as a problem only in very
> special situations (and I haven't seen any such yet).  You can use the C
> implementation in such special cases.
So can I read from this that there will be a C core to speech-dispatcher
(or whatever TTS api takes its place)? I am slightly confused on this as
else where there seemed to be discussion that it would be entirely
python, with bindings for other languages to access the python modules,
so still requiring python. The cases I am thinking of are minimal
installations/live distros, where a user might be using speakup, brltty,
yasr, etc, and would not even have an X-windows installation of any sort
available. I do accept that with increasing capacities of USB memory
sticks and external hard disks coming down in price the need for the
absolute minimal is less now, but I still would like to have the minimal
still available should I need it. I am discussing this here as a user as
I will admit I know very little of the internals of speech-dispatcher. I
do though understand the advantages of using languages such as python
though, as from doing some projects of my own I feel have been
simplified by using python.

>     MW> As for the performance issues,
>
> [...]
>
> Please note that Speech Dispatcher is not a supercomputing application,
> so there are usually quite different performance issues than optimizing
> low-level loops and function calls.
I do realise that, and may be the suggestions I made of improving
performance wouldn't be the ones for speech dispatcher, it was purely to
show that python can be improved in the performance area with very
little change to source code, so performance might not be the greatest
concern.

>     MW> In the case of orca, I feel the need for python isn't such a
>     MW> problem as gnome requires so much space anyway that adding
>     MW> python to the list of dependencies won't make a big difference.
>
> Whether you like it or not it's only matter of time when you won't be
> able to work seriously on common computers without a graphical user
> environment (I haven't already been for several years).  It's more
> important to make it well accessible than to care about advanced
> features in special situations.
My feeling is that speech-dispatcher isn't a gnome applictation, or a
GUI application and is used outside those contexts so possibly those
other applications should be considered in the design. I just meant by
the comment that a gnome application (so orca) isn't affected so much by
having to include python as a dependency as might the affect be on
something like brltty or YASR.

Michael Whapples

> Regards,
>
> Milan Zamazal