using eclipse

[Ryan M.]

Well, almost complete success here.

The only flaw I have right now is the fact that rm doesn't exist under windows, so a clean doesn't actually work. But hey... that's minor.

[prograid]

Originally Posted by Ryan M.

Well, almost complete success here.

The only flaw I have right now is the fact that rm doesn't exist under windows, so a clean doesn't actually work. But hey... that's minor.

What problems are you having with clean? It doesn't actually clean for me. What is does when I say clean is delete the entire configuration folder (the one that says either FRC Debug or FRC Release) and then rebuild it. I'm using windows too.

[Ryan M.]

Originally Posted by prograid

What problems are you having with clean? It doesn't actually clean for me. What is does when I say clean is delete the entire configuration folder (the one that says either FRC Debug or FRC Release) and then rebuild it. I'm using windows too.

It gets to "CreateProces( (null), rm files.etc)" near the beginning of the clean, then says "error, file does not exist." I'm assuming it's talking about rm, because I know that doesn't exist under Windows.

P.S. I'm working on creating an NSIS for this. It should be a quick way to install MCC18, MinGW, and Eclipse with all the necessary plugins. The one thing I'm going to check on is if I'm able to distribute MCC18 as part of the package. It;s probably prohibited. If anyone knows, I'd be extremely tickled if you'd share.

[prograid]

Originally Posted by dasRastel

yea -- no dice when adding it to the $PATH

It seems like it may be a problem with your installation of the linker. Try running the linker at the linux equivalent of command prompt. If that works see what the differences are between that and the Eclipse console. If running it from the command line doesn't work, then it's not a problem with eclipse.

Originally Posted by Ryan M.

It gets to "CreateProces( (null), rm files.etc)" near the beginning of the clean, then says "error, file does not exist." I'm assuming it's talking about rm, because I know that doesn't exist under Windows.

Well, you were right. I edited the plugin so that it now uses del /f instead of rm. I checked it out and there are no more of the CreateProcess(...) errors. The new version is in the attachment below.

This fix only applies to windows not linux. It will stop working if you use this version of the plugin for linux.

By the way, if you want the new plugin to work, delete the old one and any projects created with that plugin. You can copy the source code somewhere else, but project files need to go. Also delete the following

* /configuration/org.eclipse.osgi/manifests (this is a directory)
* /configuration/org.eclipse.osgi/.bundledata (this is a file)
* /configuration/org.eclipse.osgi/.state (this is a file)

Then restart eclipse.

Originally Posted by Ryan M.

P.S. I'm working on creating an NSIS for this. It should be a quick way to install MCC18, MinGW, and Eclipse with all the necessary plugins. The one thing I'm going to check on is if I'm able to distribute MCC18 as part of the package. It maybe prohibited. If anyone knows, I'd be extremely tickled if you'd share.

I don't think Microchip would appreciate you distributing an installer for their proprietary compiler. If it were possible, you should make the compiler install MinGW ask the user for the location of the C18 installer (or check to see if it is already installed), run that installer, and then install eclipse.

[Ryan M.]

Originally Posted by prograid

Well, you were right. I edited the plugin so that it now uses del /f instead of rm. I checked it out and there are no more of the CreateProcess(...) errors. The new version is in the attachment below.

Cool. Thanks.

Originally Posted by prograid

I don't think Microchip would appreciate you distributing an installer for their proprietary compiler. If it were possible, you should make the compiler install MinGW ask the user for the location of the C18 installer (or check to see if it is already installed), run that installer, and then install eclipse.

That's what I'm figuring. Fortunately, I don't even have to worry about where the compiler is installed... I think. It does set the PATH up, correct?

[Ryan M.]

Ok... almost got my installer working. It installs everything, sets up the PATH, but I get:

**** Full rebuild of configuration FRC Debug for project FIRST ****

Build error
(Exec error:Launching failed)

when I try to build.

What exactly fails there? This is a on a seperate laptop (not the desktop I've had working with Eclipse before).

-- EDIT --
There's a nice size savings with the installer. 111MB versus the 250MB of files it installs.

[prograid]

Originally Posted by Ryan M.

Ok... almost got my installer working. It installs everything, sets up the PATH, but I get:

**** Full rebuild of configuration FRC Debug for project FIRST ****

Build error
(Exec error:Launching failed)

when I try to build.

What exactly fails there? This is a on a seperate laptop (not the desktop I've had working with Eclipse before).

-- EDIT --
There's a nice size savings with the installer. 111MB versus the 250MB of files it installed.

The only thing that I can think of at the moment is that you didn't add the path for MinGW, it should be C:\MinGW\bin.
I don't know if this should matter, but did you install MinGW, Eclipse, CDT, and then my plugin in that order?

[Ryan M.]

Originally Posted by prograid

The only thing that I can think of at the moment is that you didn't add the path for MinGW, it should be C:MinGWbin.
I don't know if this should matter, but did you install MinGW, Eclipse, CDT, and then my plugin in that order?

Ok... I restarted after discovering a mysterious error which wouldn't let me open Firefox (something about "A previous installation must complete before this program may be run." Never seen it before, other than installers.)

Anyways, I was able to completely build (with no errors) our source code from this year.

If anyone would like to do some testing for me, there are two ways to get the installer (it's a 111MB download, be advised)

1. Download it using a BitTorrent client (I prefer Azureus) by opening up the torrent file at http://traherom.homeunix.net:6969/to...02FA 7B8E35A7. If you download it this way please leave your BitTorrent client open so others can benefit from it!
2. If that BitTorrent is giving you problems, you can download it directly from me at http://traherom.homeunix.net/setup.exe. The BitTorrent will probably be faster, as my max upload speed is only 50 k/b (usually more like 30 ).

Other miscellaneous notes:

• To use it, you must install/have installed the MCC18 compiler.
• I still can't get a normal C++ project to build. ("Exec error: launching failed") If anyone doesn't have the problem, solves the problem, or even just has an idea of what the problem is... could you tell?
• The thing seems "finicky," as stated by prograid. Not sure why.
• It required a restart (of the comp, not just Eclipse) for it to get to work for me. Try that. Don't know why, may have been some weird memory corruption thing my laptop does. It has issues.
• You must still do the steps prograid gives to create a new project. Just start after the installation is complete. I may add a readme of how to make a new project into the installer later.
• I think the license has a misspelling. Oh well. It just tells you that if you shoot yourself in the foot, it's your fault.

-- EDIT --
Sorry it the server was done there for a bit. My router has been really flakey lately.

[prograid]

Originally Posted by Ryan M.

To use it, you must install/have installed the MCC18 compiler (included with MPLAB).

Actually, it's a separate installation. You should have gotten it off of the Microchip FTP site at the beginning of the build season. You can also use the stripped down version included in the CD that came in the kit of parts, but it won't compile camera code.

[dasRatsel]

oh dear -- ur right -- my problem is somewhat more intricate than i thought -- i suppose i'll just cross my fingers and hope somebody from the adam-bots sees this post::

i ran this in the command prompt:

mplink /Files/FIRST/code/falcon/18f8520user.lkr /aINHX32 /m Mapfile.map /o FRCfalcon.hex encoder.o gyro.o ifi_startup.o ifi_utilities.o main.o pid.o printf_lib.o robot.o user_routines.o user_routines_fast.o   /l/opt/mcc18/lib /Files/FIRST/code/falcon/FRC_alltimers.lib

and got this ->

bash-2.05b# mplink /Files/FIRST/code/falcon/18f8520user.lkr /aINHX32 /m Mapfile.map /o FRCfalcon.hex encoder.o gyro.o ifi_startup.o ifi_utilities.o main.o pid.o printf_lib.o robot.o user_routines.o user_routines_fast.o   /l "/opt/mcc18/lib" /Files/FIRST/code/falcon/FRC_alltimers.lib
Microchip Linker
Linux WINE wrapper version CVS-HEAD
Copyright (c) 2004 Adambots-Live
http://adambots-live.sourceforge.net

Done!

/Files/FIRST/code/falcon/18f8520user.lkr: line 1: //: is a directory
/Files/FIRST/code/falcon/18f8520user.lkr: line 2: //: is a directory
/Files/FIRST/code/falcon/18f8520user.lkr: line 3: //: is a directory
: command not foundalcon/18f8520user.lkr: line 4:
/Files/FIRST/code/falcon/18f8520user.lkr: line 5: LIBPATH: command not found
: command not foundalcon/18f8520user.lkr: line 6:
/Files/FIRST/code/falcon/18f8520user.lkr: line 7: //FILES: No such file or directory
/Files/FIRST/code/falcon/18f8520user.lkr: line 8: FILES: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 9: FILES: command not found
: command not foundalcon/18f8520user.lkr: line 10:
/Files/FIRST/code/falcon/18f8520user.lkr: line 11: CODEPAGE: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 12: CODEPAGE: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 13: CODEPAGE: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 14: CODEPAGE: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 15: CODEPAGE: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 16: CODEPAGE: command not found
: command not foundalcon/18f8520user.lkr: line 17:
/Files/FIRST/code/falcon/18f8520user.lkr: line 18: ACCESSBANK: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 19: DATABANK: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 20: DATABANK: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 21: DATABANK: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 22: DATABANK: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 23: DATABANK: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 24: DATABANK: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 25: DATABANK: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 26: DATABANK: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 27: DATABANK: command not found
/Files/FIRST/code/falcon/18f8520user.lkr: line 28: ACCESSBANK: command not found
: command not foundalcon/18f8520user.lkr: line 29:
/Files/FIRST/code/falcon/18f8520user.lkr: line 30: SECTION: command not found
: command not foundalcon/18f8520user.lkr: line 31:
/Files/FIRST/code/falcon/18f8520user.lkr: line 32: STACK: command not found

to get this output i had to create /etc/mcc-wrapper/mcc18.conf that is just an empty dummy file

included is my .lkr file

any help would be appreciated
thanks

~Xavier

[prograid]

Originally Posted by Ryan M.

I still can't get a normal C++ project to build. ("Exec error: launching failed") If anyone doesn't have the problem, solves the problem, or even just has an idea of what the problem is... could you tell?

I just realized that when you try to create a new C++ project (either standard or managed make) the default make command is make. You have to change it to mingw32-make. If you don't want to have to change this every time you create a new project, you can add a new system variable called make and make it equal to mingw32-make.

This problem will happen whether you have used the installer from Ryan M. or not.

[Ryan M.]

Originally Posted by prograid

I just realized that when you try to create a new C++ project (either standard or managed make) the default make command is make. You have to change it to mingw32-make. If you don't want to have to change this every time you create a new project, you can add a new system variable called make and make it equal to mingw32-make.

Hm... odd. Somehow my desktop worked fine without that. Must have picked something else up.

Thanks for the info.

[WizardOfAz]

I've been using Eclipse for java in my day job for a couple of years. Finally got tired of the minimal functionality of MPLAB as an IDE and looked for some help getting going with Eclipse for FRC code. Also, we've started using CVS on the team to share and backup code, so of course Eclipse is quite nice for this with its integrated CVS support. So I was very happy to find this thread with lots of good info.

I got a basic setup working. However, in recent years we've had the practice on Team 1011 of using multiple MPLAB mcp files to produce hex files for different purposes. Basically the two motivations for this are:
(1) to produce a version that works with either the FRC or EDURC. We use the cheaper EDURC on some of our prototype bots and want to have the same code base for an EDURC and a FRC version.
(2) to produce derivative loadable codes that share most .c and .h files, for example, to make a diagnostic code distinct from a functional code.

For the first case, using different build configurations works out pretty well. As with the mcp files under MPLAB, I can use the different build configurations to define a preprocessor variable EDURC that can be used where needed in the code (not many places) when the behavior or I/O usages needs to be different between the EDURC and FRC versions. I also have different include and search paths to get the .h and .lib files appropriate for either the EDURC or the FRC.

I haven't quite figured out how to accomplish the second case. The closest I got was to make a main project and then have other derivative projects that refer to the main one for most files. I can get as far as sharing .h files and .lib files, by making search paths in the derivative project that refer to the main project. But I don't see how to get it to find .o files needed by the derivative project that are located in the main project. And you won't want all the .o files - just the ones that are required to resolve link time references. E.g. you don't want the .o file that contains "main" from the main project if you have a definition for "main" in the derivative project.

One way to fix this would be to have the main project create a .lib of it's .o files, then have the derivative project refer to that .lib. I guess this is pretty straight forward enhancement to the makefile generator. I'm going to look at doing that - unless somebody out there has better advice.

Thanks
Bill Bennett
CRUSH 1011

[prograid]

Wow! I though this thread was dead. Well , anyway, I'm not exactly sure what you mean when you say

[QUOTE author="WizardofAz"](2) to produce derivative loadable codes that share most .c and .h files, for example, to make a diagnostic code distinct from a functional code.[/quote]

I think you mean that you're using dependent projects to edit only certain files, but I'm not really sure.

[WizardOfAz]

Yeah, the derivative projects thing is mostly about changing only a few of the files, and not wanting the changes to be done to the files in the main code that will run the real bot. The best example is the diagnostic load we wrote in 2004. We had a four wheel steered drive system, lots of controls on the OI, and wanted a diag code that exercised all that stuff but would not be extra code in the competition load. So we forked off a few files into a diagnostic program, and inherited all the rest. This is what doesn't seem exactly simple in the eclipse way of doing this. Although it's easy with eclipse in java, not so with this C setup.

Bill