[omniORB] Two questions

[Duncan Grisby]

On Wednesday 3 September, Serguei Kolos wrote:

> 1. The C++ proxy and skeleton code - does it depend on platform/compiler ?
>     In the other words, is there a guarantee that code generated for the 
>     same IDL will be the same on all the supported platforms?

The same code is generated regardless of your platform.

> 2. In the omniNames implementation there are tree useful classes:
>     ReadersWritersLock,
>     ReaderLock,
>     WriterLock,
>     From my point of view they are generic enough to be added to the 
> omnithread library.
>     What do you think about that?

They might be useful in a library, but I think omnithread should
remain small, with only the basic facilities required for concurrency
control. A separate library for things like reader/writer locks would
be sensible.

An unfortunate thing is that the implementations of ReadersWritersLock
etc are under the GPL license, not the LGPL, so you can't use them in
closed-source code. Any library with them in should use new
implementations to avoid the GPL restrictions.

Cheers,

Duncan.

-- 
 -- Duncan Grisby         --
  -- duncan at grisby.org     --
   -- http://www.grisby.org --