Re: [MiNT] wind_xget()

["Arnaud BERCEGEAY" <arnaud.bercegeay@free.fr>]

Hello,

>  Good, I just checked in the changes, WF_FIRSTAREAXYWH should not be
> implemented in XaAES.

Gemlib is updated too... as well as my little test proggy
http://arnaud.bercegeay.free.fr/gemlib/tstxget2.tgz

> > The missing piece is a way to detect the availability of this function.  
> > I
> > propose appl_getinfo(AES_WINDOW(=11)). bit 9-15 of out1 seems to be  
> > unused
> > at the moment, so bit 9 may be used to indicate that WF_FIRSTAREAXYWH is
> > implemented.
>
>  Okie, sounds find by me. Should we do so?




> > About identification of AES, if i sum-up, we have:
> >
> > #define APGI_VERSION 96
> > appl_getinfo(APGI_VERSION, &min_version, &maj_version, &arch, &status)
> >
> > and for APGI_AESINFO (value = 0x5800 ???), we can think of an alternate
> > version of appl_getinfo()... something like:
>
> Yes, value of APGI_AESINFO should be agreed on.. suggestions?
>
> > appl_getinfo_str( short mode, char *str1, char *str2, char *str3, char
> > *str4);
> > which set
> > aes_addrin[0] = str1;
> > aes_addrin[1] = str2;
> > aes_addrin[2] = str3;
> > aes_addrin[3] = str4;
> > and aes_control[2]=1 (intead of 5) to tell that nothing is expected in
> > intout[1..4].
>
>  Ok, as a way to detect which aes we can use intout[1/2] to contain a 4-
> letter ID, like 'MyAES' or 'XaAES' in addition to the above definition.

'MyAES' : 5 letters
'XaAES' : 5 letters too ;)

that don't fit in a long (32 bits) integer.

If we need an AES_ID, we could add this to appl_getinfo_str():

char aes_name[64], aes_version[64], aes_id[8];
appl_getinfo_str( APGI_VERSION, aes_name, aes_version, aes_id, NULL);

BTW, i'm not sure we need to detect if the AES running is xaaes or myaes.  
If the application want to use a modern feature, then the application just  
have to check if the feature is available by calling the corresponding  
appl_getinfo() mode.

the name of the AES is only usefull if the application want to display  
it... and for that purpose a simple string is the best choice.

> > So the AES shall check the number of entries for addrin and intout in
> > aes_control[] to check which appl_getinfo is called. We even can go
> > further and use the same mode (APGI_VERSION) for both:
> >
> > short min_version, maj_version, arch, status;
> > char aes_name[64], aes_version[64];
> >
> > appl_getinfo(APGI_VERSION, &min_version, &maj_version, &arch, &status)
> > appl_getinfo_str( APGI_VERSION, aes_name, aes_version, NULL, NULL);
>
>  Yes, that sounds excellent to me :-)
>
> > where
> > aes_name will be filled in with "OurAES"
> > aes_version will be filled in with "1.02b (25 dec 2004)"
> >
> > opinions ?
>
>  I like it. With the addition of AES ID (which is a fixed-lenght AES ID)
> in intout[1/2] for
> appl_getinfo(APGI_VERSION, aes_name, aes_version, NULL, NULL);
> I really like this idea :)

What about aes_id (length fixed to 8 chars) in the 4th parameter of  
appl_getinfo_str() ?

best regards,
Arnaud.