AVR Freaks

The download link tests fine for me with Internet Explorer. Are you clicking the link or trying to right-mouse click it and "Save Target As...". Using the regular mouse button works fine for me, I suspect trying to save with the right-mouse button will lead to the problem you mentioned.

If you get the same results with explorer, it would be interesting to see what the index.php file contains. That would not be a file from my computer.

I see the URL for the download link is "https://www.avrfreaks.net/index.p..."

Can you check if the file named index.php is really the zip file. If it is, then a file renaming error is occuring that is not occuring on my system.

If the index.php is really a .php file, then for some reason the server is giving you the .php code file rather than presenting for download the actual .zip file. This seems less likely than the first senario.

It appears that you did get the proper .zip file, but for some reason your browser is not properly renaming it. Change the name of the file to serial_lcd-20071230.zip and then you should be able to unzip it.

Well, I know the steps in how it should work correctly, but I don't know why it happened to you. The name of the server file that replies with the HTTP headers to download the file is named index.php. However, as part of the HTTP headers, the name of the file it should be stored as is given. Why it did not store under that name on your sysem is not clear to me.

Thanks for the info, Johan. This was my first upload. However, the naming did work correctly on IE 7 / Vista SP1rc1.

Thanks for the report, Ross, and you're quite welcome!

Thanks for the feedback, cpluscon. I'll fix the typos and cut-and-paste errors you notice. I supplied a Makefile for GCC that compiles the C and C++ code and generates .map and .lss files. That Makefile should link fine to crttn2313.o

As for the constructors, you're referring to the empty constructors, I assume. That's a good / quick optimization idea of eliminating the empty constructors.

Edit: I commented out the empty constructors, but don't get any difference in the binary code size.

cpluscon, that's a good theory about getting rid of all constructors.
In general, I like constructors. However, in this embedded application, when I want the objects to live in data space, I there is some benefit to have control over when the initialization code is executed. Thus, the reason for using init() methods. Does that sound like a reasonable approach in this setting to you? (This is one of those areas where C++ adds flexibiliy over C, so a decision how to best design is required).

I'll check the map files. The next release will contain the map and list files for all of the compilers that I referenced in the README.

I've released a new version of the distribution. The main changes are:

- removal of C++ constructors which reduces code size for IAR & GCC
- inclusion of map and assembly output for each compiler
- addition of serial_lcd_tester which runs under Cygwin and Linux to test the serial_lcd firmware

Edit: I've deleted the posted code since I've moved hosting of the project to http://www.avrcode.com/serial_lcd/

A third part of this "challenge" could be examining the program and data memory overhead of bundling a third version of the firmware which used real C++ objects which are initialized with ports and pins identifiers and the objects store their hardware parameters. This would be useful if one wanted to create a C++ library that can be easily included and the functionality of the object being set as parameters to the init function.

Thanks for the feedback, Johan. I posted before the difference was in the C code was calling non-inlined functions. I've since fixed that and have improved the encapsulation of the C++ version significantly without increasing the code size.

I'll get a new version posted today.

Attached is the 20080104 release. The main changes are further C++ encapsulations, use of GCC's OS_main attribute, and the full-object C++ version now having full LCD initialization parameters. Those initiailization paramaters (like the LedPwm parameters) allow one to have multiple objects of that class attached to different ports.

Edit: I've deleted the posted code since I've moved hosting of the project to http://www.avrcode.com/serial_lcd/

I just finished adding Codevision compatibility. The current version is at http://www.avrcode.com/serial_lcd/
The flash usage in order of size is:

With Lee's mention of a work-around for emitting #asm and also my writing of an ISR macro for CVAVR, there are now only two #ifdef's for CVAVR in serial_lcd.c file.

Edit: maybe Lee's work-around for #asm will also work for emitting #pragma which would then eliminate the need for the _Pragma operator. Still, I think there is merit in supporting standards so maybe Pavel will add _Pragma().

Lee: I posted version 20080320 -- I'd welcome any improvements that you can make for CV's code size.