Sim Zacks wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The data in the longblob field might be text, which could be causing the confusion. For example, when I look at the data in the longblob field, I see /n for a newline and when I look at the bytea it is 012.
That's right - newline is ASCII 10 (or 12 in octal).
I can only tell you what happened in the client end, in terms of corruption. I am using the Thunderbord client. When I clicked on a message, it didn't show the data and when I looked at the headers, it was just a big mess. I'm guessing that somehow the newlines didn't work and the headers and message were overlaid on top of each other.
Well that might be a problem with dmail's setup rather than the database. I think headers are restricted to ASCII only (the body is a different matter). The best bet is to be certain whether the database is to blame.
Find a problem entry, dump that one row to a file from MySQL, do the same from PostgreSQL and also from the midpoint in your Python code doing the transfer. Then use a hex editor / dumper (e.g. "hexdump -C" on linux) to see what bytes differ in the files.
-- Richard Huxton Archonet Ltd
