Compiler 3.222 Cant run Program from Ver 3.219

[Andreas]

Hi Friends,

I have a quit nasty Problem with the Compiler 3.219 so I decided
to change to Version 3.222 and now I have much bigger troubles.

The program which is well tested with 3.219 and runs perfect is
now with the new version just starting runs a few milliseconds and then
resets to zero and execution stops !!!!


The problem I had is the following:

The initial values are:
testcalc = 3000
RAMBuffer1[3] = 52


CODE Begin

testcalc = RAMBuffer1[3] - 0x30;
RPage = RPage + testcalc + 100;

CODE End

This will give me a correct Result !
Result : RPAGE = 3400

But using this:

CODE Begin

RPage = Rpage + ((RAMBuffer1[3] - 0x30) + 100);

CODE End

Result : RPAGE = 3144

This is wrong !!


I hope somebody out there can help me !

Whatsgoing wrong with new version ???

best regards

Andreas

[PCM programmer]

If you want us to test it, you need to post a complete test program,
with the #include statement for the PIC, and variable declarations
so we can see the data types that you're using. Also show the
#fuses statement, etc.

[Mark]

3000 + (52-0x30) + 100 = 3104 so none of the values you gave are correct. Please post a small but working program that demonstrates the problem. This eliminates all the guess work.

[Andreas]

Hi ,

Sorry but the whole Code is very large allready so I just made a snippnet.

The follwoing line had a misstake:

RPage = RPAGE + testcalc * 100; Result is correct 3400 decimal !

Also this line had the same misstake:

RPage = RPAGE +((RAMBuffer1[3] - 0x30) * 100); this gives the wrong result 3144 !

Thanks both of You to reply so fast and sorry for the misstake !

Processor used : 18F8720

declare for testcalc: int16
declare for RPAge: int16
declare for RAMBuffer1[1056] : int8 ; Is the Receive Input Buffer

best regards
Andreas

[Mark]

Casting problem. 4*100 = 400 but since RAMBuffer1[3] is 8 bit then 400 or 0x190 becomes 0x90 or 144. You will need to case RAMBuffer1 to an int16.

[Andreas]

Hi Mark,

Thank for Your reply, sounds to me very logical now !

The main reason why I was so uncertain is , that one line above I made this
Statement:

RPage = ((RAMBuffer1[2]) * 1000);
And this worked fine !

So am I right that there is a difference in the behavior of C depending on
the syntax ?

This line : RPage = RPage + ((RAMBuffer1[2]) * 1000); would not work ?

Thanks again for Your answer

best regards
Andreas

[Ttelmah]

Not really.
The reason the latter example gave the expected result, is that '1000', is automatically a long, since it won't fit in a normal integer. The compiler will always use the 'highest' type involved when dealing with a pair of numbers, to determine the arithmetic to use. If you used:

RPage = ((RAMBuffer1[2]) * 100L);

Note the added 'L', this too would work as you expect, since it forces the constant to be treated as a 'long', and again forces the higher type to be used. If you look through some of my past posts about type casting, I have suggested that whenever you want to force this behaviour, using the 'L' marker is a good way of ensuring the required behaviour.

Best Wishes