From: Donald V. L. <do...@sb...> - 2004-09-26 19:54:36
|
Hi Ray, I went forward with an attempt to compile the SF download. It crashed when it could not find the /usr/src/rtai directory. 73, Don... -----Original Message----- From: emc...@li... [mailto:emc...@li...] On Behalf Of Donald V. Lemke Sent: Sunday, September 26, 2004 1:33 PM To: emc...@li... Subject: RE: [Emc-users] motion control cards Hi Ray, Ok. I went through the changes, but the result was the same as my first attempt. 20 files. Does that mean the compile is incomplete? 73, Don... -----Original Message----- From: emc...@li... [mailto:emc...@li...] On Behalf Of re...@up... Sent: Sunday, September 26, 2004 10:57 AM To: emc...@li... Subject: RE: [Emc-users] motion control cards Hi Don Thanks for editing my post. I'm working from memory here and made some rather serious omits and assumptions in the first tome. My first assumption is that emcold will contain a copy of the existing emc and rcslib from /usr/local after you have unpacked the emc_rcslib.src.tar.bz2 that is found there. The new directory named emcdevelop is where you would put the copy of the CVS at SourceForge repository. After that checkout I refer to the stuff in this directory as the repository. I should refer to this as the copy of rep or just emcdevelop. If emcold is a copy of /usr/local/emc and emcdevelop is a copy of the sf repository try the following. Grab emcdevelop/emc/scripts and move it to emcold/emc/scripts. Take packupsrc from emcold/emc/scripts and copy it to emcold. Run it and it should build a tarball. (It complains about gawk but that should not make a difference for us right now) This will produce a tarball named something like emc-3.4-26.src.tgz. The date stuff in the middle will be different. I really screwed up here. Make a third directory wherever you're working and name it emctest. I usually wind up with a bunch of these so might name it emctestold. Put the tarball in there and untar. This should give you a pristine set of sources for both emc and rcslib. It is this set that you want to compile as step A. To do that edit compile_bditng so that lines 11 and 12 read > >setenv NONREALTIMEPLAT 'linux_rtai' >setenv REALTIMEPLAT 'rtai' Save the revised and run it. There is a command that I often use to save a complete copy of the stuff that the make returns to the terminal. That command is [script]. You enter it before you issue ./compile_bditng. As soon as compile completes enter the command [exit] and it will save all of the terminal messages as a file named typescript in the same directory. [script] will overwrite typescript each time it's run. I believe that this procedure should get a full copy of rc's emc and rcslib if the Morphix install went according to plan. Now you should be able to continue somewhere in the original instructions about #6. Ray On Sun, 26 Sep 2004 10:01:51 -0500, emc...@li... said: > > Hi Ray, > > Below, in step 4 below, you say " that new directory". Which of >the two new directories are you refering to? I assumed you were >refering to emcold, and continued. I got 20 files, not 55. There was no >generic.run generated in emcold. > > 73, Don... > > >-----Original Message----- >From: emc...@li... >[mailto:emc...@li...] On Behalf Of >re...@up... >Sent: Sunday, September 26, 2004 7:41 AM >To: emc...@li... >Subject: Re: [Emc-users] motion control cards > > > > >Hi Jon, Donald > >Here are the steps that I use to compile with the Morphix or other BDI >distribution. I know that this takes time and fiddling but it allows >me to see where my compile problems really reside. > >In short, > (A) Recompile the old code that Paul used to build the Live or other >release. This tests the completeness of the Morphix environment. > > (B) Compile a stock copy of the current CVS repository so that I know >that it will compile on my Morphix. > (C) Edit the source files that need to be changed and compile again. > >All this is cumbersome and requires a bit of coffee and xanax but it at >least allows me to point directly to the area that is causing problems. > >One example of why I do this is that you can NOT compile a current copy >of the CVS using a stock install of bdi-220b. There is a header >package missing after the install. You will catch this during the >first step if you use my three step process. (BTW That package, >pciutils, is on the bdi disk. It just does not get put into place) > >Here is how I do it for the Morphix. > >1. Unpack the tarball in /usr/local so that the emc and rcslib sources, >used at the time of the rcxx build get put into the proper places. > >2. Make a new directory -- lately I've been working in my home >directory. That way I can work as a user rather than as root. I name >it emcold. > >3. Make a new directory -- mine is named emcdevelop. Download the sf >repository. You'll need this to compile the old stuff my way and >you'll need it to bring your emc up to date. > >4. Grab a directory out of the repository emc/scripts/generic and copy >it to the same place in emcold. Copy packupsrc from that directory >into emcold and run it. You'll get a dated tarball with all the source >that you need and none that you don't. > >5. Test compile the old code moving the tarball made in #4 into that >new directory. Unpack it [tar xzvf emc???] and you'll find a >compile_tng script. >I need to update this in the repository but for now edit that script so >that lines 11 and 12 read > >setenv NONREALTIMEPLAT 'linux_rtai' >setenv REALTIMEPLAT 'rtai' > >save your changes and run it. When it's done, look in >emc/plat/linux_rtai/bin and see that there are about 55 files. If yes >you've got a complete (not good just complete) build. Test it by >running generic.run. It's a stepper version that uses one parport and >minimill definitions. If that works then you have in place all of the >development stuff you need. Do NOT throw this out. You'll need a >directory from here or from /usr/local/rcslib for the next section. >You'll also want to copy the edited compile_bditng over. This >completes part A. > >6. Run packupsrc on the sourceforge stuff in emcdevelop so that you get >a dated tarball again. > >7. Make a new directory near the others -- I name it emcnew. Put the >tarball in there and unpack it. > >8. Overwrite compile_bditng with the one you edited in step 5 above. >Overwrite the entire rcslib/etc directory with the one from >/usr/local/rcslib or from the location in step 5. (You must do this in >order to preserve the def files that Paul edited for Morphix) > >9 Run compile_bditng and when it is complete look to see if you got a >complete build. If so test it by running generic.run. When you've >gotten this far, you know that the sourceforge stuff will build. (B) > >10. Edit the source files in your copy of the repository as you need >for your custom version. Packupsrc, make another new directory, >emcmine, overwrite the files like you did in #8 and compile again. >(C). > >By following this pattern, I can accurately report compile problems. I >also avoid the need to run make clean and worry about whether I've got >a good build environment. Sorry to be so pedantic but an emc build is >not a trivial thing. > >Ray > > > >On Sat, 25 Sep 2004 23:28:34 -0500, emc...@li... >said: >>Donald V. Lemke wrote: >> >>> Hi Jon, >>> >>> Ok, got allot farther, but apparently (?) bridgeportio did not >>>compile and is now not present on the system. The last make command >>>indicates this, and when strating up EMC it stops when it cannot find >>>bridgeportio. >>> >>> >>Great! We're in the same boat! I am having the same problem getting >>this to compile. I think everything made by the makefile after that >>will also be missing. And, I can't get the makefile to do more than a >>couple lines of the make process for the RT side of things. I've sent >>some questions on the emc developer's list, but haven't gotten any >>response yet. >> >>I know I'm doing it wrong (or have something incorrect in the way it >>was all set up, as there have been no significant source changes here, >>lately.) when I do these makes, I put "2>&1 | tee make.log" at the end >>of the command line. This redirects the standard error onto the >>standard output and pipes it all to the file make.log, as well as >>writing it to the screen. Then I can pick over the error messages at my > >>leisure. When I do the >> >>make clean all PLAT=linux_rtai >> >>I get : >>(78 % of file snipped) >>gcc usrmotintf.c -c \ -I/usr/local/emc/plat/linux_rtai/include >>-I/usr/local/emc/plat/linux_rtai/include -I/usr/local/rcslib/plat/ >>linux_rtai/include \ >> -g -O2 -I/usr/src/linux-2.4.25-adeos/include -DLINUX -Dlinux_rtai >>-DPLATNAME=\"linux_rtai\" -L/usr/lib -Dl >>inux_2_4 -I/usr/src/rtai/include -Dlinux_2_4_9 -Dlinux -DHAVE_LIBDL >>-DLINUX_KERNEL_2_4 -fstrict-aliasing - >>I/usr/include/g++-2/ -DNO_RTL -DHAVE_RTAI -DHAVE_FSYNC -o >>/usr/local/emc/plat/linux_rtai/lib/usrmotintf.o ) >>usrmotintf.c:131:22: rtai_shm.h: No such file or directory >>usrmotintf.c: In function `usrmotInit': >>usrmotintf.c:1099: warning: assignment makes pointer from integer >>without a cast >>make[1]: *** [/usr/local/emc/plat/linux_rtai/lib/usrmotintf.o] Error 1 >>(more stuff snipped) >> >>But that is the part where I see the first serious error in the non-RT >>compile. >> >>Jon >> >> >> >>------------------------------------------------------- >>This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 >>Project Admins to receive an Apple iPod Mini FREE for your judgement on > >>who ports your project to Linux PPC the best. Sponsored by IBM. >>Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php >>_______________________________________________ >>Emc-users mailing list >>Emc...@li... >>https://lists.sourceforge.net/lists/listinfo/emc-users >> >> > > >------------------------------------------------------- >This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 >Project Admins to receive an Apple iPod Mini FREE for your judgement on >who ports your project to Linux PPC the best. Sponsored by IBM. >Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php >_______________________________________________ >Emc-users mailing list >Emc...@li... >https://lists.sourceforge.net/lists/listinfo/emc-users > > > >------------------------------------------------------- >This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 >Project Admins to receive an Apple iPod Mini FREE for your judgement on >who ports your project to Linux PPC the best. Sponsored by IBM. >Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php >_______________________________________________ >Emc-users mailing list >Emc...@li... >https://lists.sourceforge.net/lists/listinfo/emc-users > > ------------------------------------------------------- This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 Project Admins to receive an Apple iPod Mini FREE for your judgement on who ports your project to Linux PPC the best. Sponsored by IBM. Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php _______________________________________________ Emc-users mailing list Emc...@li... https://lists.sourceforge.net/lists/listinfo/emc-users ------------------------------------------------------- This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 Project Admins to receive an Apple iPod Mini FREE for your judgement on who ports your project to Linux PPC the best. Sponsored by IBM. Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php _______________________________________________ Emc-users mailing list Emc...@li... https://lists.sourceforge.net/lists/listinfo/emc-users |