[Tinyos-host-mote-wg] [Tinyos-2.0wg] 08/03/2005 Telecon Notes

[Kristin Wright]

-------------- next part --------------

Notes from 08/03/2005 TinyOS 2.0 Working Group 

Phil Bounnadonna (Arched Rock), David Culler (Arched Rock), 
David Gay (Intel Research, Berkeley), Ben Greenstein (UCLA),
Vlado Handziski (TU Berlin), Cory Sharp (Moteiv), Martin Turon (Crossbow),
Kristin Wright (UCB)

Agenda
-------
- tools
- serial
- timer TEP update
- other

tools
------
- dgay: have built up the 2.x tools directory so you can do configure,
  make, make install what will go in a tinyos tools rpm; still need to
  incorporate cory's motelist and setup; shouldn't take too long.
- dgay: one issue: would like to preface tool names with tos; requires
  haking make scripts to use new names; since we have to figure out how to 
  do this in 1.x -- symbolic links? 
- vlado: what about generic tools? 
- dgay: not the ones that we don't provide (uisp, ncc, msp430tools*)
  if you look at 2.x 
- cory: no particular thoughts on how to go about switching
- dgay: when you  tinyos 1.1.15 which includes the new tools make; okay?
- kwright & cory: okay
- dgay: nesdoc
  - added support to nessc to output XML tags nesdoc can use
    to parse what it needs-- got one answer on mail for doing XML pass
  - could release nesc 1.2 without nesdoc
  - kwright: nesdoc is very useful
  - dgay: split nesdoc out and release as a separate package; leave the current
    one in and do a later release based on an updated 2.x; right answer: detect
    that you're using generic components and die 
  - culler: clear that it's a hole to be filled in later

serial
------
- ben: java and mote end are working 
- culler: bw are we able to get?
- ben: no idea; essentially another rewrite to move to the new interfaces;
  left no time for analysis; don't know improvement in reliability or 
  bw yet (no)
- culler: cost of double-buffering across hop still an unknown
- dgyay: is serial okay for pre2
- ben: yes

timer tep
-----------
- cory: looked very briefly at timer tep. will make it a priority this week.
- cory: noticed martin commited code for hw resource code (thanks)
- culler: revision proposal: 
  - tep follows the layering of the code so it would start
    with the raw HPL, then counters, then alarms, then timers built on 
    counters, then high-level services; appendix shows binding to resources
- culler: does that seem like the right thing to do?
- dgay: sounds reasonable. will be prefaced by a big-picture 
- cory: will be done next week
- martin: experience: he had a dilemma on whether to use generics lower 
  or introduce at HAL; found that generics worked nicely at HPL ; 
  especially with different-widthed timers. 
  - cory: found the same thing in msp430; worked out really nicely.

Other:
---------
- Message_t:
- ben: as currently implemented it is
  typedefed, but to bridge from one radio to another the header must be
  at least a s large as the largest header; how is that going to work?
- dgay: 
  - phil working on that; believe from discussion with Phil
    that the header would be a character array which is the max possible
    data one is the maximum payload footer; use GetData() in interface to
    get pointer to data; header would be right-justified against payload; 
    footer left-justified against tail of payload 
  - phil will have a write-up; defer discussion until we see that or we
    can email him
- Pre2 deadline
  - still 8/12
-------------- next part --------------
_______________________________________________
Tinyos-2.0wg mailing list
Tinyos-2.0wg at Mail.Millennium.Berkeley.EDU
https://mail.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-2.0wg