On 18 Feb 2005, at 14:24, Thomas Sundberg wrote:

I would approach this problem by isolating the where the problem occurs.
I.e. Remove, comment out, half of you insert queries and see if the problem
still exists. If it does, then remove yet another half of the remaining part
leaving only 1/4 of the original problem to solve. This would give the
solution in approximately log(n) times where n is number of parts you can
remove. I.e. If you have 100 different inserts, you would find the correct
one in about 5 tries.
This is of course the brute method you always fall back to when you don't
even know where the problem exists. Having 100 000 different inserts would
increase the number of tries to 11 or 12.

Well, that's exactly the kind of thing I'm trying to avoid. The hassle is that I have 1.7 million records spread over 26 tables, with hundreds of warnings generated when I import. I can't believe that there's no way of having these reported - does no-one value their data? Why doesn't log-warnings do anything? Should I report it as a bug? I really don't want to spend days trawling through/slicing up 3Gb of dump files when a trivial settings change might just show me them all in one go. As part of the effort to get more detail, I've switched to using single instead of extended inserts (also because many of my records hit 30k each). Could you imagine working with a compiler that only ever told you 'there was an error'?


Ultimately I want to tune the output of the dumps and fix any data problems at source so I can do transfers smoothly - my 4.1 DB is my dev version, and I'm wanting to transfer regular dumps to make sure I'm working with up-to-date stuff. That should be a matter of not much more than file transfer and import time, not several days to manually track down errors every time.

Marcus
--
Marcus Bointon
Synchromedia Limited: Putting you in the picture
[EMAIL PROTECTED] | http://www.synchromedia.co.uk


-- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]



Reply via email to