[Scummvm-devel] ScummVM 0.10.0 release soon
Kostas Nakos
knakos at phys.uoa.gr
Sun May 20 00:06:14 CEST 2007
Hello All,
Eugene Sandulenko wrote:
> o AGI engine. It is quite solid, and we will get a SoC student to work
> on it soon, only thing which prevents me from enabling it is adding
> fallback detection. I will try to do it in next few days
I would just like to point out 3 things with respect to AGI, from the
perspective of a user with limited keyboard input (e.g. phones). The
predictive input dialog gets the work (read, typing of commands) done in
this case, but:
1) Currently, the words.tok file has to be piped through a couple of
perl script to produce the list required at runtime. This strikes me as
a bit of a cruel user requirement, that is to have perl installed on the
host system and to prepare the dictionaries beforehand. Are we
considering moving the functionality from perl to the agi startup, say
when detecting that no dictionary has been created, to do so at runtime?
(I know it must be complicated to actually save given the allowed
savepaths and so on, but it can be compiled & kept in memory)
2) It would be good if the predictive dialog handled keys 0..9 and
possibly * and # directly (this is trivial)
3) I still think that the user should be able to bring up the pred.
input dialog by pressing a key, instead of moving the simulated cursor
down to the command line and clicking. There should be a better way to
do this, other than hack-ishly injecting a mousemove & buttondown/up events.
I could be able to code all these in, but I have a rather busy schedule
for at least one week more (and in the worst case, 2 weeks).
Are these things to be taken into consideration? It would be great if
they were implemented before the release but, at least for the WinCE
port, it is not like the agi engine is not working, just not doing so in
the best way.
> o WinCE port. Kostas also released an updated version recently. I hope
> it works well
Yeah, the svn build is looking good, only two things to deal with before
a stable release. I hope I can get another svn build out during the bug
fixing period. After that I could tackle the above mentioned things in
AGI's predictive input, provided there's free time and that we want them in.
Cheers,
K.
More information about the Scummvm-devel
mailing list