[EMAIL PROTECTED] ([EMAIL PROTECTED]) wrote: > when specifying ClientPut/RemoveLocalKey=true, does fred remove the stored > key (chk) BEFORE or AFTER trying an insert?
If you removed the key after insertion, then the insertion will not have worked, because the key will be gone. So that wouldn't make much sense. > experiments with build 618 (btw - the new layout (bullets, small title frame, > typo) sucks in my opinion) revealed, that a ClientPut succeeds with > RemoveLocalKey=true, even if that key *already existed*! this leaves me to > think that the *old* key is removed before the *new* key is stored, leading > to no key collision. Well, yeah. That's kind of the point. > the IMHO correct way would be to remove the *new* key *right after* it has > been inserted, and if such a key already exsits when trying the insert, > returning a KeyCollision! But then you'd be deleting keys from your data store and not replacing them. Given that the point of the RemoveLocalKey option is to improve propagation of your keys, deleting from your data store afterward seems to be a step backward. If your goal is to remove all traces of the key from your data store after insertion to "hide your tracks", then I think you should consider stopping your node and running "rm -rf store_*" or "DELTREE store_*". And make sure you're using a transient node for this purpose, not a permanent one. -- Greg Wooledge | "Truth belongs to everybody." [EMAIL PROTECTED] | - The Red Hot Chili Peppers http://wooledge.org/~greg/ |
msg02159/pgp00000.pgp
Description: PGP signature
