[Tinyos Core WG] Preparing for the RPM

[Joe Polastre]

Actually, not completely true.  If #include is the first thing, and
then the nesdoc documentation comes after but before the
configuration{}, implementation{} etc lines, then it will generate
correctly.  We had to deal with this in Boomerang, and the easiest fix
was to put #include first.

-Joe

On 10/30/06, Kevin Klues <klueska at gmail.com> wrote:
> In components that #include some file before the configuration or the
> module begins, nesdoc doesn't seem to generate documentation for the
> component.  Moving the #include inside the implementation section
> fixes this, but it shouldn't need to be this way.
>
> Kevin
>
> On 10/30/06, Vlado Handziski <vlado.handziski at gmail.com> wrote:
> > On 10/30/06, Philip Levis <pal at cs.stanford.edu> wrote:
> > > On Oct 29, 2006, at 10:55 PM, Kevin Klues wrote:
> > >
> > > > TEP3 says that all directory names should be lower case, but all of
> > > > the applications have directory names in upper case.  I'm not sure if
> > > > the TEP or the directoriy names should be changed, but I thought I
> > > > would at least point out this discrepency.
> > >
> > > I think it might be useful to make the application distinction.
> > >
> > >
> > > >
> > > > Also, TEP3 says that type names for structs should be all lower case
> > > > separated by underscores and ending in '_t'.  Until now, we have
> > > > always followed the convention that message structures are defined
> > > > like "RadioCountMsg"  or "OscilloscopeMsg".  Do we really want to
> > > > change these to be radio_count_msg_t and oscilloscope_msg_t?
> > >
> > > Hm. This is a funny one; it's because the structs are mapped into
> > > Java classes, and historically the class names have been the same as
> > > the struct names.
> > >
> > > Overall, I think it's better to keep the code the way it is and
> > > adjust TEP 3 accordingly.
> > >
> > >
> >
> > But consistency also counts. By simply changing the MIG lines in the app
> > Makefiles, we can have the structs using _t while keeping the java names
> > unchanged. Since not all C message structs have their java equivalents at
> > the moment, it seems like less work this way.
> >
> > Vlado
> >
>
>
> --
> ~Kevin
> _______________________________________________
> 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
>