Jun 24, 2008 - But then you have to have the table on a mod 256 boundary (not a killer condition, C or ASM) and you have the 'page' address of the table always. But about the USI; instead of using the USI and flipping the bits, it might be beneficial to write a bit-banged software UART/SPI/TWI/1-wire/whatever bus. Simple Serial Communications With AVR libc., Serial.println() and Serial.read() methods. At the same time AVR Libc has proven printf(), puts() and getchar().

I've moved up to the latest release of WinAVR. Unmodified my old code compiles as it did with the previous version. If I cause a header file to be changed, the compiler then reports a problem with fdevopen(). Samsung Yp S5 Software Upgrade. Firstly:./main.c:211: warning: passing arg 1 of `fdevopen' from incompatible pointer type./main.c:211: warning: passing arg 2 of `fdevopen' from incompatible pointer type then it complains there are too many arguments. Has something happened to fdevopen? If I remove the 3rd argument (the '0') the error disappears. If I then compile again with no further changes, it's all OK again.

Any ideas anyone? Besides of Mr. Download Matrix 3d Jewelry Design Software.

Noname cemox obviously ignoring the smiley after the 'RTFM', let me tell you that it takes literally *dozens* of hours to eventually write all that f*ing documentation, so please accept that the only thing that then will really p*** off those who wrote it is to see people completely ignoring it, and then crying for help. Sorry for the offending language. If you want to see that written in more words with a not-so-offending language, follow the advise to read the ``intro' thread: Otherwise, everyone who is able to help will eventually turn away, and avoid talking to the ignorant people at all. Hi Cliff, I'm not using printf() or sprintf() at all at the moment, but I know how useful it can be for debugging. I have written my own functions for displaying hex, decimal, and scientific format (you know, 1.23E-5, that sort of thing - a breeze in C compared with PIC assembler, I can tell you!) I was just curious, having been reading the manual about how fdevopen() would cause malloc() to be pulled in as well, as to whether the sprintf() option (once again, only for debugging) would involve a bit less baggage. Yup, personally, though the fdevopen()/printf() stuff is nice, I'm a luddite and tend to do my own 'raw' hex/decimal output routines and then just have a uart_send_str() type function (oh and a uart_send_str_P() for my PSTR()s ) so as to not drag in any baggage like malloc() or the fully featured printf() but sometimes (just for debug) I 'give in' and use a bit of sprintf() to prepare the buffer full of stuff to be sent, I often wrap it all in #ifdef DEBUG so it's easy to switch it all out later with/without a -D at compile time. These problems with version switches of avrgcc ist the only thing I hate on avrgcc.

If the changes really creates some benefit, like this uart thing it is ok. But often there are some cosmetical changes, things that become deprecated which are only replaced with another named thing with the same functionality, like outport which changes in the past from outport to outp to port=. If you use big projects like NutOS you get hundreds of errors and no benefit from the new names. But you need hours to fix it up. So please, avrgcc developers, do not change things that need no change. Please only change it when this is necessary for new functionality. Or better _add_ new functions/definitions and keep the old ones. Korg M1 Vst Crackle.

Thank you Thorsten. Apart from it being a bit pointless to reply to a one-year old thread, think about it: *often*? There was just that *single* deprecation of sbi/cbi/inb/outb, and it has been announced a whopping 5 years before these old (and rather pointless) macros have eventually been phased out.