Brecht Vermeulen writes:
Brecht> I don't know if anyone is using the -f corba functionality of Brecht> smidump, but as indicated in the source-file dump-corba.c :
Brecht> /* XXX I think we need to include INDEX columns here XXX */
Brecht> it is indeed needed (as far as I know and as far as I can find Brecht> in the JIDM documents, otherwise you can't create rows when an Brecht> index column is non-accessible or read-only) to add the index Brecht> columns always to the create_xxx functions in the interface Brecht> SmiEntryFactory : SNMPMgmt::GenericFactory {
Brecht> which are used to create rows in tables.
Brecht> So, in attachment you can find a patch which solves this. (for Brecht> libsmi-0.3.0).
Thanks. I do not have the spec handy, but it just makes sense to add all INDEX columns as parameters. I have checked the patch into the CVS. Regarding the usage of the -f corba option: I did not receive many bug reports for the CORBA backend and since it is unlikely that I wrote bug-free code, I doubt that many people use the CORBA backend. In fact, my prime motivation of writing it was to understand how the JIDM translation works (and as part of that exercise, I have send quite a few bug reports to the JIDM folks). And since I found quite a few bugs in the C802 draft version, I was asking myself how many people have ever implemented/used these JIDM translations...
Brecht> It's the first time I write some code with libsmi, so I hope Brecht> it's correct :-).
It looks good to me. I change the string comparison since you can in principle use an INDEX element which is imported from another module which has the same name as one of the table columns (pretty unlikely but legal). Libsmi guarantees you that you get the same SmiNode for the same object and thus one can just compare the pointer values.
Anyway, feel free to send more feedback about the JIDM CORBA backend and I would certainly be interested whether you are planning to use this for any production tools or just for academic purposes.
/js