#3181 printf double vsnprintf() fix, malloc, va_end

[knifter]

These are the changes mentioned in #3181

[stickbreaker]

I've just thought of something, This print function is used for many different subsystems. I think some of them may actually have cases where write() can fail. instead of returning the number of characters to be written, actually return the true number.

int actualLen = write((uint8_t*)temp,len);
...
return actualLen;

[knifter]

Ill add/fix that

[stickbreaker]

don't do it this way, the buffer has already been freed, free() will overwrite the content of temp with the delete chain links.

[knifter]

oops... absolutely right, a sec

[stickbreaker]

you could:

if(temp != loc_buf){
     len = write((uint8_t*)temp,len);  
     free(temp);
}
else len=write((uint8_t*)temp,len);
return len;

[knifter]

I also want to write(temp) if temp == loc_buf
e: right, too soon. But see commit

[stickbreaker]

@knifter Found one more at line 162:

size_t Print::print(struct tm * timeinfo, const char * format)
{
    const char * f = format;
    if(!f){
        f = "%c";
    }
    char buf[64];
    size_t written = strftime(buf, 64, f, timeinfo);
    print(buf);
    return written;
}

Instead of return written; change it to return print(buf);
That is the only other function that was not returning actual characters written.

Thanks for doing this. Every little bit helps!

[me-no-dev]

@stickbreaker that is unrelated though :)