Re: [Simulavr-devel] Version and doc fixes

[Joel Sherrill]

Weddington, Eric wrote:
>  
>
>   
> > -----Original Message-----
> > From: Joel Sherrill [mailto:address@hidden 
> > Sent: Wednesday, March 11, 2009 3:56 PM
> > To: Weddington, Eric
> > Cc: address@hidden
> > Subject: Re: [Simulavr-devel] Version and doc fixes
> >
> > Do you want to add stub sections which compare other
> > AVR simulators to this one?
> >
> > X Comparison to Other AVR Simulators
> >
> > X.1 simulavr (C version)
> >
> > X.2 avrtest
> >     
>
> Since these 2 will probably be dumped in the very near future, I don't see a 
> reason for the comparison.
>
> Joel, I don't know if you're subscribed to the avr-gcc-list mailing list, but I 
> posted a question on that list this morning about any potential 
> backwards-compatibility with simulavr. I got 5 immediate rejections of simulavr 
> (nobody could get it to work) and nobody (yet) cares about 
> backwards-compatibility.
>  
>   
Looking at that thread, I would say I read Tristan correctly.  :)

I personally rejected using avrtest for RTEMS because it didn't support gdb.

So http://lists.gnu.org/archive/html/avr-gcc-list/2009-03/msg00112.html
seems to sum it all up:

==========================
> IMO it is absolutely required that a simulator can run as gdb server 
> for avr-gdb.
>
> Stand alone simulators like avrova seem not to have this 
> functionality. I just skimmed the avrora site and saw no note pointing 
> in that direction, maybe I middes it.
>
> Some Nice-To-Have features:
>
> -- run in gdb server mode
> -- code breakpoints
> -- data breakpoints
> -- breakpoint on SP to detect stack overrun/underrun (hard because SP 
> is two SFRs)
> -- query ticks since some milestone
> -- logging/dumping (what instruction is being executed, what GPRs, 
> SREG-Bit is changing, etc)
> -- can simulate some internal peripheral that triggers IRQ, so that 
> ISRs can be debugged and tests be written for ISRs
>
> Georg-Johann Lay
>   
==========================

I can't argue with his list.  I would put IRQs higher since
I know for RTEMS you can test nearly everything with just
a debug console and clock tick.

So I would say that puts the priority on running the gcc
test suite on it.
> > X.3 Avrora
> >     
>
> That might be an interesting comparison, but hopefully in the future we can be 
> feature compatibile with avrora, minus the RF support.
>
> Overall, I don't see a great need now to do comparisons. I'd like to put this 
> off as a future task.
>
>   
OK.  I didn't realize the state of the other two.
>   
> > > The down side to this is that it seems I need to change the 
> > >       
> > file name of doc/simulavrxx.texinfo to doc/simulavr.texinfo 
> > so the filename of the pdf is simulavr.pdf. Do we really care 
> > about keeping the CVS history for this file? If not, I could 
> > just delete the one file, and add the new filename. 
> > Personally I don't think it matters.  There isn't any critical
> > history is there?
> >     
>
> It's just documentation, not code.
>   
Move it.  Full steam ahead. :)
> Eric
>   


--
Joel Sherrill, Ph.D.             Director of Research & Development
address@hidden        On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
  Support Available             (256) 722-9985