Jump to content
OpenSplice DDS Forum


  • Content count

  • Joined

  • Last visited

About simonmcqueen

  • Rank
    Advanced Member

Contact Methods

  • Website URL

Profile Information

  • Gender
    Not Telling
  • Company
  1. simonmcqueen

    find/discover on Topic (isocpp)

    Hi zickedi, Right now there are still some areas of the ISO C++ spec not implemented. You can see the most up-to-date list of what is and isn't supported here in the release notes. This on-line version is (of course) more current than that which shipped in the Community Edition distribution in July. WRT 'find/discovery' in particular please see this section. Cheers, Simon.
  2. simonmcqueen

    Cyclic include definition from idl

    I think you mean: <TopicType>_DCPS.hpp: #include "<TopicType>Dcps_impl.h [...] ... Yes - it is known. It is 'inherited' from the legacy idlpp code generation. Compare: m@plutus:~/worgit/ospli/examples/dcps/HelloWorld/cpp/standalone> find . -name '*.h' |xargs grep ccpp_HelloWorldData.h ./HelloWorldDataDcps_impl.h:#include "ccpp_HelloWorldData.h" ./HelloWorldDataSplDcps.h:#include "ccpp_HelloWorldData.h" sm@plutus:~/worgit/ospli/examples/dcps/HelloWorld/cpp/standalone> cat ccpp_HelloWorldData.h |grep Dcps_ #include "HelloWorldDataDcps_impl.h" sm@plutus:~/worgit/ospli/examples/dcps/HelloWorld/cpp/standalone> As it's not useful to link any subset of the *.cpp files it actually makes sense on any number of levels for us to just generate only one header (.hpp) and one .cpp file per IDL file, and that's the eventual goal for the ISO C++ mapping. If you want to watch for the fix in the release notes the issue ID is OSPL-130.
  3. simonmcqueen

    IDL forward references

    Hi egoforth, That IDL is illegal. An incomplete type (like Bar) can't be used as a member even if you were to forward declare it. You can only legally use a forward declared incomplete type *sequence* as a member. As James says, right now we don't support even that though, although it is on the way. See http://dev.opensplice.org/cgi-bin/bugzilla3/show_bug.cgi?id=41 We are not planning on supporting the code that you have there though AFAIK as it doesn't comply to the language specification.
  4. simonmcqueen

    Newbie questions--Pure Java? SimD for Java?

    Hi jmlineb, You can find the latest specification here: http://www.omg.org/spec/DDS-PSM-Cxx/Current The usage documentation can be found in the distribution or here http://dev.opensplice.org/releases/downloads/docs/isocpp/html/index.html You can probably answer the second question yourself by comparing the 'before' against the 'after'.
  5. simonmcqueen

    idlpp and paths

    Hi Gidi. Use 'forward slashes'.
  6. simonmcqueen

    Problem running as non-administrator on Win7

    Hi Tobbe, I can remember fixing a bug regarding the ACLs we were specifying when creating Global shared objects that are integral to our shared memory database on Windows that I think had this effect. Basically we had to relax the restrictions from what we had previously else Windows prevented unprivileged users getting at them. Unfortunately for you the fix went out in OpenSplice 5.6 or thereabouts. If I were you, I'd download an evaluation of 6.2 from http://www.prismtech.com/opensplice and re-test with that. If it works - that's probably what it is and you could look into support, wait for the next OSS release, or have a punt debugging it yourself. If not: the fault may well indeed be yours and/or this is something new we've not seen previously...
  7. simonmcqueen

    problem with os_versionString()

    Hi xoquoxxotil, The makefiles that build OpenSplice define VERSION when we compile the source via the compiler command line. It is defined from the environment value $PACKAGE_VERSION like: -DVERSION=\"$(PACKAGE_VERSION)\" See, for example, https://github.com/O...2.6-default.mak This environment variable can be set by sourcing release_info/RELEASE when compiling up from a source checkout, or etc/RELEASEINFO if you are working with a distribution. What is the nature of your problem with this function?
  8. simonmcqueen

    OpenSlice DDS test with 2 computer

    Do you mean the OpenSplice daemon (spliced) ? If so and you are using shared memory communication then the answers are: yes & no respectively. See docs/pdf/OpenSplice_Deployment.pdf and OpenSplice_Tutorial_C.pdf
  9. simonmcqueen

    How to avoid re-reading samples from durable topic

    Hi Brian, Have you looked at a subscriber instance dispose (following a publisher unregister_instance in conjunction with autodispose_unregistered_instances = FALSE) ?
  10. simonmcqueen

    How to analyse GDS memory?

    Hi syamtsn, The configured database size is the total size of the shared memory segment that will be allocated on this node. The size used will not be affected by how many nodes there are, or the configured sizes of their shared memory.
  11. simonmcqueen

    OpenSplice DDS (32 bit) on Windows 7 (64 bit)

    Hi Orhan, Generally speaking: yes, although YMMV with 5.5.1.
  12. simonmcqueen

    IDL: sequence of sequence

    Hi Changsuk, This is a known issue with the OpenSplice community edition. Please see http://dev.opensplice.org/cgi-bin/bugzilla3/show_bug.cgi?id=49
  13. simonmcqueen

    How to make a simple chat application with opensplice

    Hi adhye_alor, You're in luck! In the distribution, the directory examples/dcps/standalone/Java/Tutorial contains an example simple chat application using OpenSplice.
  14. simonmcqueen

    Order of messages at subscriber

    Hi Florian. You've made messageID part of your instance key in your IDL. This means each 'message' is an instance. Try taking messageID out of the key defintion and read up about instances vs topics.
  15. simonmcqueen

    IDL generating C# files crashes

    Hi guys, I can confirm this is a C# specific bug in idlpp. If you want to watch for a fix in the release notes the issue # is OSPL-885. Can't tell you when this will be fixed in the Community Edition right now I'm afraid but in the meantime the workaround suggested above will not only work but is in fact much better practice than the original IDL anyway. Thanks for reporting this.