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

Re: akka status



Salam Haydar,

Actually it is not really Acon vs Akka as much as Acon *and* Akka.

Acon is good if you need an approximation of bidi in your text screen.
The thing is, what you'll obtain will be very buggy, screen dependent and
therefore unadapted for production use, as well as incompatible with the
UAX#9 because bidi on a console is "mathematically" impossible. On the
other hand, Acon is great for decrypting a few pages using Lynx, or type
occasional non-arabized commands in an Arabic environment. IOW Acon is
really good at offering a minimal Arabic support in "hostile"
environment:)
Akka on the other hand simply discards bidi support. It is possible in
theory to offer some support at the input level, but I have given up the
idea because it would mean in practice redoing all the keyboard
management done at the kernel level, so it would be much easier to
integrate a few callbacks in the input drivers in the kernel than to do
it in Akka). What you win is accuracy (what you enter is what will be
stored), compatibilty with libraries which support reverse bidi,
increased compatibilty with the console-tools, and remove all the
drawbacks of the often complex but not perfectly implemented bidi
versions out there. What you lose is support for bidirectional texts,
i.e. text that will mix "significant" portions of both Latin and
Arabic. Note that numbers, math, and non-breakable sequences of Latin
text inside Arabic text or non-breakable sequences of Arabic text inside
Latin text such as isolated words for example are possible in
monodirectional texts in theory, so Akka supports them in practice. In
order not to make entrance of reverse sequences in monodirectional texts
a real pain in the toe, a "block cursor" mode (a special insertion mode)
has been added. Another tradeof is the approximative readability in non
Arabized environments (e.g. reading a latin file path without any special
support on an RTL oriented console).
So both probably complete each other.

Now, to answer you about the status of Akka. I do believe that Akka is
almost complete functionally (at least as long as we didn't have users
expressing strong feelings about the need for some given functionalities
I hadn't thought about). Depending on when exactly you need it, I think
I should be able to commit in order to complete the few functions
required (essentially linking the CORBA calls to the internal calls)
since it's almost done already. After that it'll be bug-fixing only.
On the other hand, making it user-friendly is something that you'd have
to undertake by yourself because I'm not sure I'll be able to do it in
the next couple of months. What is actually needed to make it user
friendly is more or less 1) packaging 2) writing a few default
configuration scripts in order to initialize it (such as setting up the
console fonts using the console tools for example, chosing which consoles
will be monitored, the keyboard maps used...). Configuration is machine
dependent but experience taught me that people won't use it if it's not
ready to use without typing more than "install" followed by Enter;)

Salaam,
Chahine