[Tinyos Core WG] toolchain rationale
Jan Beutel
j.beutel at ieee.org
Sat Nov 18 04:18:25 PST 2006
> > this is a snippet from boomerangs ncc where the absolute TOSDIR was
> > removed by an environment test. something similar needs to
> be done in
> > nescc-mig
> >
> > # Configuration
> > #$TOSDIR = "/opt/tinyos-2.x/tos";
> > $TOSDIR = $ENV{"TOSDIR"} if defined($ENV{"TOSDIR"}); $nescc =
> > "/opt/moteiv/usr/bin/nescc"; $tossim = 0;
>
> No. The environment test is, and has always been, there. The
> only effect of commenting out the absolute path is that
> things will fail if TOSDIR isn't defined (except if you use
> the -tosdir option). As for mig, it tells nescc-mig to use
> ncc rather than nescc, so it follows whatever ncc does
> (earlier versions called ncc directly, so the effect was the same).
>
that may be the case but as you can see if you compare boomerang ncc with
the CVS HEAD in tos2 people muck with the files (for whatever reason). and
there are a lot more executables that need to be configured/prefixed
correctly that make the story even worse. so my argument is not at all about
ncc and so but about the whole toolchain that _just is complicated_!
to make things clear: i commented out the line moteiv put in...
> As I said before: tosdir is and has always been relocatable.
> The argument always seems to boil down to whether having a
> default is a good idea (the discussion usually starts like
> this one - people don't seem to check documentation or source
> code before complaining). The other relevant question is
> where to get the location - currently it's an env var or a
> command line switch, but other choices could be considered
> (registry on windows, configuration files in $HOME, etc).
no hard feelings david... that is why i named this thread "toolchain
rationale" in the beginning. i was curious (i) if such a rational was
existent and (ii) if people are adhearing to it. seems like the answers are
(i) "mostly" and (ii) "some".
jan
More information about the Tinyos-2.0wg
mailing list