> > I think we discussed this last year and decided that it would be a bad > idea to use those names because Oracle's use of them is not exactly > compatible with our error codes and messages. SQLCODE in particular is > not compatible at all --- it's an integer in Oracle, isn't it?
There is more incompatibilities to Oracle. SQLERRM is function on Oracle, only if you use it without parametr, returns current message error. SQLCODE is really integer. But it's only names. There is no problem change it. > > IIRC we had put off solving this problem until we decided what to do > with RAISE. There really needs to be some changes in RAISE to allow it > to raise a specific error code rather than always P0001, but exactly > what is still undecided. I didn't know it. But for my work is SQLERRM more important. I have more constraints on tables and I need detect which which constraints raise exception. The possibility EXCEPTION WITH OTHERS is nice, but not too much usefull because I have not possibility get some informations about except. > > Some other problems with your patch: no documentation, and not in > diff -c format. Plain diff patches are never acceptable because > it's too risky to apply them against files that might have changed > since you started working with them. Also, it's much easier to > deal with one patch than with a separate diff for each file. > (diff -c -r between an original and a modified directory is one > good way to produce a useful patch.) > I am not sure, I able create documentation - my english is poor. I will change diff's format and send patch again. Thank you Pavel ---------------------------(end of broadcast)--------------------------- TIP 9: the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match