This is great work!
I checked the blockchain file size before I updated and after. This trimmed 4.075977% off of the blockchain on my computer. It was actually more because I hadn't synced for a day or two before I updated and resynced the blockchain. BBR already synced faster than the other CN coins and this will just make it that much faster and more usable. I can't wait to see what future developments you come up with crypto_zoidberg. Advances in the implementation of the Cryptonote technology like this are the reason that BBR IS the BEST Cryptonote coin. You have made more significant changes to the original codebase than any other implementation of CN. It's only a matter of time before that becomes a well known fact within the crypto community and BBR takes its rightful place as the leader among Cryptonote coins.
You deserve some kind of recognition for your hard work so I'm donating another 500 BBR to your dev fund. BBR transaction# 682e5ea5c373bc8f877189d3b0632c580a384135f8818fd53d6ca5fab417db4c
Keep up the good work!
Blockchain with release "bbr-win-x64-v0.2.0.37(ceecfdf)":
Blockchain after updating to version "bbr-win-x64-v0.3.0.38(109f49c)":
I feel that i need to clarify a little here.
First - as i said before - pruning transactions will make significant effect for those part of blockchain that have pretty much transactions in it. Early blockchain consist mostly from coinbase transactions that is not affected(coinbase transaction doesn't have ring signatures). But anyway, there more that 8000 usual transactions in first 70000 blocks.
Second,
blockchain.bin is not simple blockchain. When daemon receive blocks with transactions it build from this data special database in memory, there tons of helpful information that is used to verify blocks and transactions and to handle currency rules: global outputs index (that is used to refering in transactions inputs ), transactions database, some information related with each block(such as cumulative difficulty, scratchpad offeset, block cumulative size, emitted coins, etc), blocks by id database, spent keyimages, finally scratchpad.
All this information is serizlised (dumped) into this file (blockchain.bin), so it is obvious that there some huge data overhead. So actually summary blocks + transactions size is not so big, hard to say exact numbers, x times less. And this is another reason why you see not much difference - since ringsignatures is actually ballast it is not used in other data model, and after pruning it data model stay almost the same.
Main benefit from this feature is decrease traffic usage(actuall data amount that transfered between syncronizing nodes under checkpoints get significant decrease), and some times (when synchronizing reduced by bandwidth) - speed up synchronyzing.