sprintf format error

[SeeCwriter]

When a variable is declared as an int, the following sprint() statement fails to compile with a format error:

[SeeCwriter]

"sprint" is a typo, it should be sprintf.

[Ttelmah]

Do you possibly have a #type declaration in your code?.

Remember a default 'int' is unsigned, while %d prints a signed variable. Normally though I'd not expect that to complain, but if int had been re-declared, so it was an int16 (for example), then the printf would complain.

Just tried it, and if I have:

#type int=16

then it complains exactly as you describe.

You then have to use %ld

[SeeCwriter]

I have a number of typedef's, but I did not redefine 'int'.

I even tried format %u, and it still failed. It doesn't make sense.

[Ttelmah]

Not typedef's #type.

This changes the default size of a variable.

With int re-declared as an int16, then the %ld format has to be used.

Do a search in your code for #type. There will be one in the chip include file, so look carefully at what they all say.

[SeeCwriter]

I've searched all the files in PICC\Devices and in PICC\Drivers and I could find no #type statements. I also searched all my project files, and no #type.

But I'm beginning to think there is something else going on. The compiler is acting differently. It no longer generates a list of compiler errors, it just lists one error at a time. Fix that error and recompile, and it lists one error...etc. And the errors I'm getting now don't make sense, such as the sprintf() format error. I tried %u, %d, %ld, %lu, and they all generate a format error if the variable is declared as 'int'.

So I change the variable to 'byte', and that error goes away, but the next one is "Function used but not defined." But it is defined. I have a whole file that is nothing but function prototypes that is included right after the chip header file. Nevertheless, I move the prototype into the same file as the call. Still get the error. Move the prototype to just before the call, still get the error. I finally rewrite the function into a #define statement. Now the error is gone, but a new one shows up, and it's the next line of code: Function used but not defined. Comment out that call, and the next line is flagged as an undefined error. This is all bogus. I think the project is corrupt or there is a problem with the new compiler version, v5.046.

[PCM programmer]

In your previous thread, I point out that the CCS modified stack does
have #type statements in it, and 'int' is redefined as 16 bits:
http://www.ccsinfo.com/forum/viewtopic.php?t=53963&start=2&highlight=type

[SeeCwriter]

I did not remember that. However, I have wrapped all TCP/IP stack statements in #ifdef's and have removed TCP/IP from the project. I also renamed the tcpip directory created under my project directory just to make sure no stack files were being accessed.

That will present a problem in the future when I try to enable Ethernet capability and have to go through the program re-declaring many variables.

[Ttelmah]

First, start with a much smaller (literally ten line) program, and verify that things work correctly.

Then look at #device=ANSI. This makes the default integer size int16, so %d will print an int16. Makes the variable sizes match what the ethernet stack will expect.

Then consider if you might have something really simple, like a #define that is affecting your int definition.

On your 'error' behaviour, there is a tick box in the project options 'show all errors'. If this is unticked, only the first error will be shown. From the command line, this is the -E option. It does suggest that some project options have been changed. Look at everything here carefully. Perhaps a global define?.

[SeeCwriter]

Even wrapping all the Ethernet includes and configurations in #ifdef's to remove them from the program did not solve the problem. My last resort was to start over.

I created a new project with no Ethernet capability, and copied over all my non-Ethernet related code to the new project. And now all the weirdness is gone, I can build an executable, and it's running on my hardware.