[AL] Keyboard hander stuff

[Jared S. <Jar...@li...>]

Is there a way to get something like a 'Ctrl - Alt - Enter' combo past the
allegro keyboard handler and to the OS, without shutting down the handler? =
(Linux version of allegro)
Thanks,
Jared
Jar...@li...
>From <all...@ca...> Wed Sep 27 05:07:00 2000
Received: from finch-post-11.mail.demon.net [194.217.242.39] by canvaslink.com with ESMTP
  (SMTPD32-6.00) id A8B31D5501CE; Wed, 27 Sep 2000 05:06:59 -0400
Received: from lwithers.demon.co.uk ([194.222.80.1])
	by finch-post-11.mail.demon.net with esmtp (Exim 2.12 #1)
	id 13eDAz-0000V7-0B
	for al...@ca...; Wed, 27 Sep 2000 09:06:54 +0000
Message-ID: <oOr...@lw...>
Date: Tue, 26 Sep 2000 08:48:17 +0100
To: al...@ca...
From: Laurence Withers <lwi...@lw...>
References: <32F...@ww...>
In-Reply-To: <32F...@ww...>
MIME-Version: 1.0
X-Mailer: Turnpike Integrated Version 5.01 M <WCcCG$mmNL32sYSf46vBeC80CR>
Subject: Re: [AL] Lock functions and variables in classes
Precedence: bulk
Sender: all...@ca...
Reply-To: al...@ca...
X-UIDL: 268079491
Status: O
Content-Length: 2794
Lines: 70

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

In reply to Czamai, Martin (Czamai, Martin <M.C...@pe...>):

>my problem is, that I want to lock functions like a interrupt service
>routine for some piece of hardware in the constructor of my class MyClass.
>This doesn't work since  MyClass::MyMethod_end is no member of MyClass,
>which is generated by the LOCK_FUNCTION macro.
>Any idea how I can fix this problem??
>Thanks in advance

OK, this is quite a difficult problem to solve. Basically, Allegro locks
functions by using a dummy function, declared after the function to
lock, to get its end address. When you write:

void my_function()
{
  /* some stuff here */
}
END_OF_FUNCTION(my_function);

Allegro creates a dummy function, perhaps my_function_end or something,
directly after your function. We now assume that when your compiled
executable is loaded into memory, my_function_end is located directly
after my_function.

To lock the memory used by my_function, Allegro needs to know its start
location (simple; that is just &my_function), and its length. Computing
the length of the function requires another symbol directly after
my_function, so the length can be found as (&my_function_end -
&my_function). Now we can lock that function (hopefully!)

In a C++ class, however, we have a slightly more complex problem: name
mangling. For instance, your method's actual function name might well be
`pXkqMyClass_MyMethod' - the C++ compiler mangles the name so that it
can support function overloading (the weird characters at the beginning
would indicate the argument types).

So, your only option is to somehow find the mangled name, and write

END_OF_FUNCTION(mangled_function_name)

after your function definition. Of course, this might not work :-) You
also have one more problem - if you are constructing an object in an
interrupt context, you are allocating memory, which is a call to a
standard library routine. You are not allowed to do that in an interrupt
context.

To be honest with you, I would use the placement new operator and write
your own constructor and object factory in C. You would have to allocate
memory in advance of the interrupt and lock it. During the interrupt,
you would have to use placement new to create a new object at that
location. The operator does not allocate new memory or call the object's
constructor, so you have to construct the object yourself.

Probably your best bet is to think of another way of doing what you want
to do.

Bye for now,
- -- 
Laurence Withers, lwi...@lw...
                http://www.lwithers.demon.co.uk/

-----BEGIN PGP SIGNATURE-----
Version: PGPsdk version 1.7.1

iQA/AwUBOdBUwXcA2AQvE3AZEQI5DQCgqystMfTgwzHEMVHTexuIRNG7HJQAoJW5
iMtwJoNWExfHsFSz5yV9PLKh
=C2Hp
-----END PGP SIGNATURE-----
>From <all...@ca...> Wed Sep 27 07:45:12 2000
Received: from mail5.svr.pol.co.uk [195.92.193.20] by canvaslink.com with ESMTP
  (SMTPD32-6.00) id ADC81D670162; Wed, 27 Sep 2000 07:45:12 -0400
Received: from modem-213.phosphorus.dialup.pol.co.uk ([62.136.14.213] helo=ollie)
	by mail5.svr.pol.co.uk with esmtp (Exim 3.13 #0)
	id 13eFe6-0003uD-00
	for al...@ca...; Wed, 27 Sep 2000 12:45:06 +0100
Message-ID: <002001c02878$074f54c0$d50e883e@ollie>
From: "Ollie" <odo...@x-...>
To: <al...@ca...>
References: <000001c0281b$62fab740$0101bfbf@f9c6f0>
Date: Wed, 27 Sep 2000 12:42:32 +0100
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 5.50.4133.2400
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400
Subject: Re: [AL] Allegro 3933 with msvc :)
Precedence: bulk
Sender: all...@ca...
Reply-To: al...@ca...
X-UIDL: 268079491
Status: O
Content-Length: 436
Lines: 14

> 2) for the direct x SDK,  the doc could state that only the lib & header
> file ( dx7libhdr.exe 1.3MB ) is needed and not the whole SDK ( 120 MB !? )

Hello Matt, do you already have the complete SDK?  if so, have a look at
this to help get set up:

http://members.nbci.com/_XMCM/LouisHowe/allmsvc.htm

it's true as far as i know though, you dont need the complete SDK, it all
depends on what you are planning to do i guess.

Ollie