On 09/01/2012 11:13 PM, Werner Donné wrote:
We will have to fall back on copying the byte array in case a temporary file couldn't be created, for whatever reason.
That's reasonable, especially if we log the fact at debug/trace level too, to encourage people to turn the feature off if it doesn't apply in their environment.
Do you propose to make this always-on, or configurable?
The treashold could be configurable.
I think it'll need to be, if nothing else to stop people from having tempfile issues.
How (if at all) does this interact with JDBC-standard BLOB support, and with large objects?
It has no effect on either.
I suspected not, but it's good to confirm. Thanks.
I have never used git, nor have I written unit tests, but I'll have a look.
If you've used any SCM like CVS, Subversion, or Perforce the principles are similar for basic use. I highly recommend it. This'll be handy:
https://help.github.com/ GitHub have spent a lot of time and effort writing documentation and tutorials for git.
Sorry to ask you to do that; I'd probably just grab the source tarball and pick through it, but I'm on a travelling holiday for the next couple of weeks and just trying to check in on the mailing list sometimes.
--
Craig Ringer