> I've used slony in the past to migrate from 7.4 to 8.0, and i plan on > using it to migrate from 8.0 to 8.1 once the 8.1 issues with slony > get hammered out. Additionally, though, I'm curious if slony can be > used to migrate from one database encoding to another. We are > currently ready to migrate our database and applications from Latin1 > encoding to UTF8 encoding. Is this something slony supports? Are > there caveats that I should be aware of? I can run pg_dump with -- > encoding=UTF8 on my Latin1 database and load it successfully into my > UTF8 database and have the 8bit Latin1 characters correctly converted > into ther UTF8 counterparts.
This falls into the category of "things not explicitly tried yet." With the fixes in 1.1.2, we now have reasonably proper support for UTF8, though that has only really been tested via copying from the same encoding into the same encoding. I'd suggest trying it out; if it works well, that would be worth adding to the documentation. If there are wrinkles, that's also something others might care to know. I'd suggest testing it out with a reasonably diverse set of tests; make sure you have: a) Examples of each special character so you know each character is translated properly; b) Be sure you test both with data that is present at subscription time (which is added via a bulk COPY) and with additional data inserted/updated *after* the subscription is set up (which is introduced via INSERT/UPDATE, which is a quite distinct mechanism from COPY). Actually, if you can pass on some relevant sample of data, I'd be more than happy to see if I can turn it into a test in the "Testbed: The New Generation" ;-) Assuming it all works, that approach would make it part of the regression tests! _______________________________________________ Slony1-general mailing list [email protected] http://gborg.postgresql.org/mailman/listinfo/slony1-general
