IMO




List

Contents
  1 - Where to post Hardware Abstraction Layer stuff
  2 - Fork the sourceforge CVS to allow for HAL
  3 - Add a documents directory for HAL
  4 - Add developers to EMC
  5 - Fix Sourceforge Lists 


1 - Where to post Hardware Abstraction Layer stuff

I like keeping the HAL thread with this group (emc-at-nist.gov subscribers) for 
a bit because it generated good discussion, sense of direction, and momentum 
recently.  

We need to dual post at least to the emc-developer list, 'cause that's where 
most of the code writers read.  Some folk that we know and will need during 
this remodel direct this list (emc-at-nist.gov) into a mailbox and seldom look 
to see what's in there.


2 - Fork the sourceforge CVS to allow for HAL

My thinking on this may be a bit different from John K's post.  He suggested 
a couple of emcmots that both get compiled and then are selectable by an ini 
variable.  My meager experience with compiles of EMC tell me that if it does 
not all work, none of it will work.  This will mean that the new emcmot and 
the early attempts to develop socket stuff to run on either side of the HAL 
will need to compile or the whole thing is broke.  

I believe that the move to abstraction layers will produce a much stronger, 
more flexible EMC.  I also believe that, for a while with this effort, near 
chaos will be our primary product.  I am really excited about the changes 
that we will make but before we have finished it we will have rethought 
everything that we do now.  

For example, Professor Hugh Jack posted a very kind report to the MatPLC list 
in which he described the discussion from EMC Monday.  One reply suggested 
that their work is very similar to the HAL that is being developed here and 
that the entire EMC could run as a module within their framework.  I am not 
capable of evaluating this though but I know that both the EMC and MatPLC 
have worked from the same shared memory paper written by Fred Proctor.  It 
may well be that these projects are moving toward each other and can share 
some work.

I'd like to see us set aside a full CVS version of the current EMC before we 
begin to build in the range of changes that we outlined during EMC Monday.  
This module should be compilable and runnable on the several platforms that 
we currently support.  It should be editable so that we can continue to fix 
bugs in it.  For a while, this will be the stable, BDI, and Sherline version. 

Right now there are three CVS modules in the repository; emc, rcslib, and 
documents.  I propose that we freeze emc and add emc-hal.  Copy a reasonably 
full set of the current emc code into emc-hal and quickly hack up emc-hal so 
that it becomes our work in progress.  In spite of steep learning, we should 
start these hacks soon, small though they may be, so that we keep momentum.  
If it seems wise, make the two emcmots here and as the HAL and its logical 
cousin grow to be the core, re-order all of the directories and files so that 
they fit the new architecture. 

If we use a complete set of emc for emc-hal we would be able to packup and 
compile from either module using the same procedures.  This would be real 
handy for those of us who are compile challenged.  


3 - Add a documents directory for HAL

Add a HAL directory in the documents module to hold descriptive material for 
the new systems.  Eventually this kind of stuff should find it's way into the 
developer handbook.  Several of you have done some great work with making the 
interactions between parts of EMC and RCSLIB clearer for all of us and these 
things need to be there.  I'll try to do some cleanup here in documents and 
create the new directory.  This will not be intended to replace the excellent 
web material that John K maintains but might archive it as he chooses.


4 - Add developers to EMC at sourceforge

We need to expand the use of the sourceforge tools available to us and widen 
the base of developers that have write access to the sourceforge repository.  
Write access will allow all of us to collaborate on the documentation as well 
as the code.

I approached my developer status with a great deal of fear that I'd make a 
complete fool of myself and destroy more than one persons good work.  My 
fears proved to be true.  More than once, Will Shackleford had to fix stuff 
I'd done wrong.  Within minutes of my new status there, I messed it up.  I do 
find that as I read a chunk of code and figure out what's going on, I can 
comment that chunk and put it back in the source for the benefit of others.  

If you are willing to become a developer, you can skim my writeup on setting 
up a sourceforge account and getting developer status in the developers 
handbook.  ( I'll post a copy of this chapter in the www.linuxcnc.org/dropbox 
as developer_notes before the day is out.)

See also http://sourceforge.net/docman/?group_id=1 for more details than 
you'll need about anything sourceforge.

Paul is touring for another week so after you have set up your sourceforge 
account you should direct your request to be added to the developer list to 
Fred or Will. 


5 - Fix Sourceforge Lists 

Right now there are four almost unused lists associated with the EMC at 
sourceforge.  They are;

 Developers (2 msgs)
 Help (5 msgs)
 Open Discussion (15 msgs)
 emc-developers (0 msgs)

The last one, emc-developers, is where this note is also going.  There are 
zero messages there because it is nothing more than a device to fan out 
messages sent to it to each member on the development team.  

To start this we should designate a developer/administrator or someone 
willing to become one (do I see a hand raised?) to serve as a kind of lists 
mom for these.  No moderating here, just someone to keep up with 
sourceforge's list abilities and see that the settings for each list make it 
do what it is supposed to do.  

Then we need to cut the existing lists to two, emc-developers and emc-users.  
Make the emc-developers list work to archive messages so that anyone can read 
them but limit posting to developers.  Move the entire subscriber list from 
emc-at-nist.gov to emc-users.  Sourceforge has a mechanism that will do this.  

Close the emc-at-nist.gov by setting up an autoresponder there that explains the 
change and directs the person attempting a post to the emc-users list at 
sourceforge.  We will need to fix a bunch of links around the www when we do 
this.

Sorry for the long winded and shotgun post.  I'm just beginning to catch up 
after NAMES.  As you respond to this please snip.

Ray


On Wednesday 07 May 2003 04:09 pm, Craig  wrote in response to John K:
> ?"Judging from the deafening silence after I posted my proposal, I don't
> think much is going to happen soon.  I personally have a steep learning
> curve to climb before I'm ready to write





Date Index | Thread Index | Back to archive index | Back to Mailing List Page

Problems or questions? Contact