[omniORB] omniidl assert error about comments after upgrade from 2.8 to 3

[Duncan Grisby]

On Thursday 16 March, Richard Gruet wrote:

> There is another issue about comments and the -k option.  Following the
> Javadoc convention, I use to put them BEFORE the objects they describe,
> whereas you follow the opposite convention in idlast.py.
> How could I manage that ? Could you make it an option (a k sub-option?
> beware, you'll lack option letters soon)

It won't be especially easy to support comments before a declaration.
The way it currently works it to attach comments to the most recently
encountered declaration. (There are currently a few surprises with
this, due to the way the parser works, but I'll fix them some time
soon.) To attach comments to declarations appearing after the comment,
the comments would have to be stored somewhere, then grabbed as a new
declaration was encountered. Not particularly hard, but code would
have to be added to quite a lot of places. I might do it, but don't
expect it any time soon.

> And finally, a suggestion:
> It'd be nice if you could catch the exceptions when calling the back-ends
> (in main.py, line #400: bemodules[i].run(tree, backends_args[i])), because
> omniidl is silent when a syntax error or other exception occurs in a
> back-end, and that makes the debugging more tricky!

The version in CVS should now report errors much better. If you are
hacking back-ends around, you can always hack main.py so it doesn't
catch any exceptions.

Cheers,

Duncan.

PS. Please can you configure your mailer so it doesn't send HTML to
the list.

-- 
 -- Duncan Grisby  \  Research Engineer  --
  -- AT&T Laboratories Cambridge          --
   -- http://www.uk.research.att.com/~dpg1 --