Mares Smart Apnea

[Jef Driesen]

On 2015-09-09 14:18, Giorgio Marzano wrote:
> I don't have access to any SMART memory dump, so I can't figure how to
> detect  the variant.

I'll send you data from some mares smart's.

BTW, when working with memory dumps, you can use the libdivecomputer 
simulator. Setup instructions are available here:

http://libdivecomputer.org/simulator.html

This allows you to test the code without needing physical access to a 
dive computer. Very convenient.

> I had only two ideas, but both of them are dirty hacks:
> - detect using the string compare in the mares_iconhd_get_model and set
> some flag

That's plan B. The problem here is that we identify devices by means of 
the model number. But we already know both the normal and apnea variant 
have the same model number 0x10. So we'll need to invent some new model 
number for the apnea variant. But if a future mares device also uses the 
same number, then we're in trouble. So if we can detect the data format 
based on the data itself, then that would be a better alternative.

> - develop a separate backend

Bad idea. Devices using the same communication protocol should use the 
same backend. We don't want duplicated code.

Jef

> G
> 
> 2015-09-09 13:41 GMT+02:00 Jef Driesen <jef at libdivecomputer.org>:
> 
>> On 2015-09-04 12:25, Giorgio Marzano wrote:
>> 
>>> Right now, I noticed a bug in the mares_iconhd_get_model function, 
>>> line
>>> 120
>>> of mares_iconhd.c, where sizeof should be actually strlen.
>>> 
>>> if (memcmp (device->version + 0x46, models[i].name, /*sizeof*/ strlen
>>> (models[i].name) - 1) == 0) {
>>> 
>> 
>> This is not a bug. The name field is a 16 byte array, not a string. 
>> The
>> initialization with a sting literal is convenient. The remainder of 
>> the
>> array will automatically get padded with zero bytes, which is exactly 
>> what
>> we need.
>> 
>> The correct fix is to add a new entry containing "Mares Smart". Then, 
>> the
>> next question is: what's the correct model number? According to the 
>> model
>> number in the memory dump (e.g. byte at offset 0), it's 0x10 or SMART:
>> 
>>                 {"Smart",       SMART},
>> +               {"Smart Apnea", SMART},
>>                 {"Icon HD",     ICONHD},
>> 
>> But then how do we distinguish the normal smart from the apnea 
>> variant?
>> That will be important, because both appear to use a different data 
>> format.
>> The strange thing here is that the normal smart already had a freedive
>> mode. The main difference is that we now not only get a summary of 
>> each
>> freedive in the session, but also a detailed profile.
>> 
>> I wonder if there is some kind of format indicator in the header that 
>> we
>> can use to distinguish between dives with the enhanced apnea format 
>> and the
>> more limited normal format.
>> 
>> Jef
>>