[Tinyos-host-mote-wg] [Tinyos-2.0wg] 07/27/2005 Telecon Notes

[Kristin Wright]

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

Present: Phil Buonnadonna (Arched Rock), David Culler (UCB, Arched Rock), 
David Gay (Intel Berkeley), Ben Greenstein (UCLA), Vlado
Handziski (TU Berlin), Martin Turon (Crossbow), Kristin Wright (UCB)

Notes from 07/27/2005 TinyOS 2.0 Working Group 

07/27/2005 Agenda
-----------------
- toolchain (common to future 1.1.x, 1.2, and 2.0)
 o platform tools (external)
 o nesC
 o tinyos tools
- pre2
 o serial
 o documentation (teps, interfaces, general)
 o code cleanup
 o packaging

07/27/2005 Discussion Notes
----------------------------------------------------------

Miscellaneous
-----------------
Martin: having some compilation problems; will send mail to dgay

TinyOS Tools & Platform Toolchains
------------------------------------
- dgay: idea is to have a common toolchain for 1.1.x, 1.2, 2.0; what
  is the status of the msp toolchain? 
- kwright: Kevin Klues (TU Berlin) built a set of rpms for linux; building
  cygwin versions -- some small tweaks necessary, but going fine. Did hit
  upon a gcc snag but not worried about it -- just haven't worked through it.
- dgay: can moteiv use these?
- vlado: moteiv uses a modified bsl script; telos has an I2C 
  switch on usb to prevent access to msp over usb; their modified code
  enables or disables switch
- dgay: is it possible to merge code into one bsl? 
- vlado: problem is that modification based on old version of bsl; this version
  works for them and they don't plan to move on. will discuss with cory; believe 
  they might be able to come to a compromise
- dgay: motelist: does tu-berlin use that?
- vlado: no we don't; will check it in what they do use. Their tool depends 
  on what version of jdk/serial library, plus there is a licensing issue with
  sun's jdk.
- dgay checked in cory's serial port stuff for 2.x
- dgay: not ideal, but ok to have different solutions, but we should have 
  the same installation story; number of steps should be reasonably small; 
  ex: pybsl, motelist should be in same package (motelist not working in linux 
  which complicates things)
- vlado: to really solve problem motelist: whenever you plug in a node, then
  we should generate an event and change permissions of user. Perhaps have a 
  user tinyos/group tinyos (change permission to tinyos/tinyos)
- dgay: have same issue for micas; have a hack that he uses
- vlado: would need to do a lot of testing. 
- dgay: have a clear README on how to deal with them; dgay, vlado, cory to 
  figure out how they deal with the issues. will send out an email to the whole
  list about impending meeting.
- dgay: nesc 1.2 alpha: nesdoc doesn't work at the moment. doesn't deal with any of the 
  new features. xml hackers? would like to do something similar but woudl have
  an xml database. more stable, more reliable. 
- culler: doesn't require deep tinyos knowledge.
- dgay: gil, cory, kamin might be obvious people to ask. 
- dgay: will send an email to tinyos-devel. 

pre2 release
--------------
- dgay: targeting aug 12 for toolchain / target for pre2 also aug 12; mica2 only, 
  other platforms can join next pre2. 
- culler: key there is getting a modern timer on the atmel; radio stack carries
  over 
- dgay: timermilli is there, what's missing is 32khz
- turon: working on that
- dgay: assuming we *wouldn't* have 32khz that in pre2
- culler: so pre2 will be a subset of the tep; have we nailed down binding between
  logical and physical resources? 
- martin: have been discussing with cory how to multiplex 32khz timer; did not
  modify tep. 
- culler: ideal thing would be if we could get people to explain how they're going
  to implement the tep on their particular platform
- culler: want to know logical (mille, 32khz) to physical resources
- dgay: has cory revised the timer tep? 
- culler: will try to work with cory on that.
- serial support
  - ben: separate framing from escaping; separate framing from dispatch; module that
    just escapes bytes to something called SerialM which implements same framing
    protocol that we had before. AT the framing layer, there's a type field and 
    also have the AM multiplexing at the messaging layer. Framing: acks, sending data
    with request for ack, how to convert to a message_t. code is checked in, haven't 
    begun testing. 
- dgay: estimate on when ready with basic functionality? 
- ben: a week from now. will keep us updated. noted that there is still one open issue
  in addressing.
  - dgay: can we defer that until post pre2? doesn't have to be final version.
  - ben: yes
- dgay: code cleanup: phil will work on when he gets back
- dgay: documentation
  - teps should be readable and accurate. (1,2,3,101,102,106, 107, 108, + oski). 
  
    - culler: any feedback from public release? 
    - dgay: a few things
  - interfaces w/proper documentation; 
  - overview of pre2 release and an explanation of how 2.0 is different than 1.0 
    (1 paragraph for each system and then point to TEPS); README in each app 
    directory as before. not worry about tutorial.)
- dgay: should wait on overview until phil has done code cleanup; 
- culler: would like to work on the overview but can't take the lead.
- kwright: will work on packaging.
- dgay: philb: are you working with imote2? 
  - philb: not clear.
  - dgay: should we talk to lama?
  - philb: good idea. 




-------------- 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