|
From: Jim M. <Jam...@Su...> - 2001-11-30 17:13:05
|
Hi Viktor, I actually have the implementation working with 0.9.2, I will get 0.9.3 and try it with that - I don't anticipate any issues here. I will look at combining the NPITool.cpp file so that there is only one. However; there has to be 2 NativeProvider.java files because the interfaces that Sun uses are more aligned with JSR 48 then what the SNIA interface currently is (I think). I'll let you know how I make out with it. Viktor Mihajlovski wrote: >Hi Jim, > >I'm really excited to hear that you are working on a WBEM services port. I >think that all your concerns are valid, and I have prepared a new NPI >release 0.9.3 which should address your issues. Indeed, the code shouldn't >cause warnings, regardless of the compiler. The SampleProvider hasn't been >touched for a while (internally we use the SBLIM providers as "samples"), >hence the "incompileability". I've fixed that and also have added a >makefile and a MOF for the sample. I'd suggest you base your work on the >new release. > >Regarding your port I would suggest that we try not to have two seperate >NPITool.cpp implementations, but rather try to handle the difference via >the method/class tables initialization. Ultimately the SNIA CIMOM should >have the same method signatures (except for the package names) and I really >would love to see that being handled by one single implementation. Let me >know what you think about it. > >Mit freundlichen Grüßen / Kind Regards > Viktor Mihajlovski > >Linux Technology Center > > IBM Laboratory Böblingen, Germany > Phone +49-7031-16-2560 > E-Mail mih...@de... > > > > Jim Marshall > <Jam...@su... To: sbl...@ww... > > cc: > Sent by: Subject: [Sblim-devel] NPI compilation issues > sblim-devel-admin@www-1 > 26.ibm.com > > > 11/29/01 08:41 PM > > > > > > >Hello, > I am looking at adding a port to the SBLIM NPI so that it will work >with the Sun WBEM Services (and our open source one), but I've hit a >couple of problems which I would like some information on. I am using >npi 0.9.2 and using the Sun 'cc' compiler (version 5.3), I've hit two >problems in npi.h, in there there is the following enum > >typedef enum { > CIM_INTEGER, > CIM_STRING, > CIM_REF, >} CIMType; > >Notice the ',' after "CIM_REF" - this causes a compile warning with the >Sun cc compiler ('Expected identifier instead of "}"') > >Additionally the METHOD_STRUCT causes lots of "String literal converted >to char* in initialization" because the 'methodName' and 'Signature' >members are declared as 'char*' and not 'const char*'. > >Lastly in the sample C provider the 'createInstance' function calls >"CIMObjectPathNew" passing the CIMInstance as the second parameter - the >Sun cc compiler complains because that function expects a char* as the >second parameter. Shouldn't this be calling >'CIMObjectPathFromCIMInstance'? > >The main question I have, is if I make the above changes, would this >adversely affect the other implementations in any way? I can work >around these, but figured I'd ask if it would be ok to modify them. > >Thanks >-Jim > >_______________________________________________ >Sblim-devel mailing list >Sbl...@ww... >http://www-124.ibm.com/developerworks/oss/mailman/listinfo/sblim-devel > > > |