I am migrating a website from Oracle 8i to 11g, and one of the snags I've run into is the character set.
The old database has both NLS_CHARACTERSET and NLS_NCHAR_CHARACTERSET set to US7ASCII. This character set seems to have disappeared in 11g. I tried making a database using Oracle's defaults, which are WE8MSWIN1252 and AL16UTF16 respectively. The data looks ok so far, but my client is worried about problems we just haven't seen yet.
My question: are there other choices for the two settings which would give us a better chance of having unmangled data than the ones Oracle chose? If it matters, he's worried about the diacritics specifically, though I don't think he has actually found any yet that look wrong.
Also, a couple of related questions:In 8i, I set the NLS_DATE_FORMAT by adding a line to the init.ora file. In my Googling I have seen suggestions to set this by writing out a pfile, editing that, then converting it back to an spfile, basically doing the same thing I used to do but with the added steps of translating from and to the binary format. This seems clunky; isn't there a better way?
Lastly, the client noticed that NLS_TIME_TZ_FORMAT has changed, not to his liking. Would I set this in the same way as NLS_DATE_FORMAT, whatever that turns out to be?
Thanks for being patient with some very newbie questions! janine -- //www.freelists.org/webpage/oracle-l
