AW: [omniORB] omniidl bug

Rene Jager renej.frog at yucom.be
Wed Apr 14 13:28:35 BST 2004


On Wed, 2004-04-14 at 11:41, Serguei Kolos wrote:
> Hi Barthel
> 
> I believe this is not an issue. I don't know why it has been working
> before,
> but normally one has to specify with the -I option all the directories
> for the 
> included files. For example C++ compiler does not include files from
> the
> current directory unless you provide it with the "-I." option in the
> command line.
> The same option should solve the problem you described, i.e. try:
> > omniidl -bcxx -I. 

if two.idl includes three.idl using quotes, and the files are in the
same directory, and the preprocessor can find two.idl, then the
preprocessor should include the file three.idl. Using <>'s the directory
should be given to the preprocessor with -I

renej

> 
> Cheers,
> Sergei
> > Hi Sergei
> > 
> >   
> > > As a result the original (short) file name appears in the preprocessor output.
> > > I believe the omniidl itself does not try to open the included files and everything 
> > > works as I desired. I believe this would not have any negative implications to the
> > > IDL compiler. What do you think ?
> > >     
> > I reported this bug also and was very interested in the fix/hack.
> > 
> > I tried it yesterday and it had some minor/major problem:
> > 
> > the directory where the root preprocessed file is opened is nomore added automatically to the search path.
> > 
> > 
> > Example:
> > 
> > /firstdir/one.idl     (does #include "../secondir/two.idl")
> > /seconddir/two.idl    (does #include "three.idl")
> > /seconddir/three.idl
> > 
> > 
> > now in two.idl the include three.idl cannot be found.
> > 
> > I assume the current "local directory path" is generated from the whole filepath to two.idl by stripping of the filename two.idl. Your fix may break this.
> > 
> > The current fix may also break make-runs for some people when using omniorb 4.0.4 in the future

worse, the CORBA spec says that the preprocessing should be conform
c/c++ preprocessing... that seems broken right now
don't take me wrong, I'd like an intuitive behavior of include in IDL ->
include in C++ very much ;-)

> > 
> > Except for the shown problem it works fine.
> > 
> > -marco
> >   
> 
> 
> ______________________________________________________________________
> _______________________________________________
> omniORB-list mailing list
> omniORB-list at omniorb-support.com
> http://www.omniorb-support.com/mailman/listinfo/omniorb-list




More information about the omniORB-list mailing list