Re: [transfer-dev] Re: clone() before delete()?

2010-08-16 Thread Mark Mandel
What happens if you try the clone? Honestly, Transfer was built with this sort of referential failure was an non-expected error, and can (apparently) cause issues. I think the easiest option - do the delete by SQL, catch the error. If it doesn't error, discard the object from the cache. Mark

[transfer-dev] Re: clone() before delete()?

2010-08-12 Thread Brian G
Mostly because there are many reasons why RI would prevent a delete so the number of checks is many and keeping them up to date might prove mistake-prone. Brian On Aug 11, 4:27 pm, Mark Mandel mark.man...@gmail.com wrote: Stupid question, why not just do a check with some sql, rather than

[transfer-dev] Re: clone() before delete()?

2010-08-11 Thread Mick Hutchinson
Hi Mark, Looking through the code for transfer one would expect objects that have been discarded not to exist in the cache. However, I ran a simple unit test overnight and it proved otherwise. As suggested I made the Transfer.isCached() method public. The test involved randomly loading 1000

[transfer-dev] Re: clone() before delete()?

2010-08-11 Thread Mick Hutchinson
Sorry... wrong thread. Not sure what happened there. Will repost. On Aug 12, 9:36 am, Mick Hutchinson hutchinso...@gmail.com wrote: Hi Mark, Looking through the code for transfer one would expect objects that have been discarded not to exist in the cache. However, I ran a simple unit test