Last week Alex merged in the changes I had made to the local 40m copy of the Real Time Code Generator. These were to add a new part, called FiltMuxMatrix, which is a matrix of filter banks, as well as fixing the filter medm generation code so the filter banks actually have working time stamps.
I checked out a new version of the CDS SVN with these changes merged in. Changes that will be added in the near future by Rolf and Alex include the addition of "tags". These are pieces in simulink which act as a bridge between two points, so you can reduce the amount of wire clutter on diagrams. Otherwise they have no real affect on the generated C code. Also the ADC/DAC channel selector and in fact the ADC/DAC parts will be changing. The MIT group has requested the channel selector be freed up for its original purpose in matlab, so Rolf is working on that.
For the time being, Alex has created a directory /rtcds on Linux1 under /home/cds. He then created softlinks to that directory on megatron, c1iscex, and allegra in the /opt directory. This was an easy way to have a shared path.
After checking out the CDS SVN, we discovered there some files missing that Alex had added to his version, but not the main branch. Alex came over to the 40m and proceeded to get all those files checked in. We then checked it out again. Changes were made to the awg, framebuilder, and nds codes and needed to be rebuilt.
Certain other file name conventions were also changed. Instead of tpchn_c1.par, tpchn_c2.par, etc, its now tpchn_c1lsc.par, tpchn_c2lsp.par, etc. The system name is included at the end of the filename, to help make it clearer what file goes with what.
This required an edit of the chnconf file, which has explicit calls to those file names. Once we edited that file, we had to reload the xinetd service which its apparently a subpart of (this can be accomplished by /etc/init.d/xinetd stop, then /etc/init.d/xinetd start).
/etc/rc.d/rc.local also had to be edited for the new model names (c1lsc, c1lsp, etc).
The daqdrc file (for the framebuilder) now parses which dcu_rate to use from the tpchn_c1lsc.par type files, so the dcu_rate 20 = 16384 lines have been removed. set gds_server has also been removed, and replaced with tpconfig "/opt/rtcds/caltech/c1/target/gds/param/testpoint.par"; from which it can get the hostname. This information is now derived from the c1SYS.mdl file.
After that Alex informed me the IOP processor needs to be running for the other models to work properly, as well as for the Framebuilder to work.
Initially there was a problem running on Megatron, because the IOP gets its timing signal from the IO chassis, and there was none connected to megatron. However, he has since modified the code so that if there's no IO chassis, the IOP processor just uses the system clock. It has been tested and runs on megatron now.