Topic: Bottlecaps 2.1 UPDATE REQUIRED - HARDFORK JULY 4 2014 to 200% Annual PoS - page 168. (Read 388610 times)

How close to release of latest update? And anyone have any idea of timeframe before back on Cryptsy?
I agree with Isawhim in that exchanges could just continue trading internally with coin already in existence as opposed to shutting it down completely but it is what it is..just hope its not off exchange for too much longer...and curious how much more this will devalue the coin too...

I like the new wallet design 

Actually that's radioactive... (Sort-of)

Might want to make the arrows matching the wallet icons, the curved ones... For the top icons. (Just my 2-cents. It would be easier to follow. More standardized/themed.)

Yea, those floating point values are no good for anything other than 3D and generalization. (No possible way to fix them, due to them being pre-calculated from LUTs in the CPUs and backed by LUTs in the OS's, which all have variations and quirks.)

Decimal and "format" or Currency is more reliable, unless you really need 13 decimals of sort-of precision. (I believe true 64-bit linux and OSX may use native double-double, converted to double, which is not the same as just a double answer.)

It's always math that screws us in the end...

Where possible, you get faster results if you do the multiple instead of the screwed-up division, where possible. EG, replace all /2 with *0.5, all /4 with *0.25, etc... or even x*(1/2) is faster than x/2, because the LUT for 1 is the largest and fastest, and the multiply function is 10,000x faster than division of a variable value, then the LUT of the fixed values. (Constants help there too. Acts like a custom LUT for you. Using div_2 = 0.5, div_3 =0.3333333333333333, div_4 = 0.25, etc... Then all your answers are the same on every machine. 100% Then you use them like this... if X * div_3 >= Y then...) {LUT = Look-up-table}

I like the sounds of the mentioned plan of attack for the updates. Sounds like some logically wise choices. However, for a future update, I would still like to see the suggestion of the DB separated into two parts. The archive of blocks, relevant to the last hard-coded block, which never has to be deleted, and only gets updated when the program becomes updated. And the separate self-repairing "new blocks", that would be the only portion we would have to delete, in the event of failure to self-repair when trying to connect to a larger block-height. Since the programs seem to choke and keep us mining on the wrong chain.

Actually, we only need the last tallest block to mine on the correct chain. The database can be fixed at a leisure, if it just got the last block hash of the tallest chain that others have already confirmed as valid. Might want to look into that idea first.

EG, I have block 85000 but the tallest chain that 3 more have confirmed was block 85150... Just give me that block, and let me build off that. WHEN the wallet fixes itself, now that it knows it is forked, it will end-up showing my block that I found 85151... eventually... as opposed to keeping us mining on the smaller chain, mining block 85001, which it KNOWS will be an orphan anyways.

That would put your wallet/system above and beyond what any other wallet/system offers...

LOL We wouldn't want anyone to think that Bottlecaps were toxic.

Better?

Mullick: Great. I will review the changes once published... I know a thing or two about portability issues (I am the author of http://ta-lib.org ).

John: I understand many relates bottlecaps to fallout, but to me (I do not play such games) the radioactive sign is giving me a weird vibe. I am hopeful that this is an optional skin. No big deal one way or the other, just my humble feedback.

Both: Thanks for the updates. Greatly appreciated.

Is this what the updated client looks like? or is that just what your client looks like?   Be cool if that was the update!

Thanks for the update. Looking forward to the fix!

Yes sorry for delayed response Im still working on this no worries

It appears to be more than a few problems.

1. Floating point math could be a issue. It can be interpreted different ways by various compilers. I am getting rid of it all.

2. Computeminwork is still not adjusted correctly. PoS blocks on occasion still fail this condition causing the forks. This could be due to the floating point math originally used but it is uncertain

3. 1 Minute stake spacing with a incredibly low maturity is not really sustainable. Its just maddness

Therefore there will be a 2 stage release for these issues

The first of which I will be temporarily having computeminwork() return minimal difficulty hopefully to stop the forks for now a well as removing all floating point math. Its only temporary until the correct function can be verified.

Stage 2 will be a major rework of Caps.

Confirms for both PoS an POW will be changed to 120 instead of 25 this will help in the future with issue like the first fork

Stake target spacing will also be changed form 1 minute to 3 minutes to make it more sustainable. The reward change to 5% will be included in this fix

More details to come. John Eden is currently working an a new design for the 1.5 wallet while I work on these changes

EDIT: Also going to look int fixing the banned peers issue it just seems to compound the problem

My understanding is Balthazar found the error in the math essentially this .016 x 24x 60 x 60 should be 4 x 60 x 60 anyhow long story short it is fixed. Confirms for mining and minting also had to be changed to 120 as according to all the Proof of Stake masters this is the lowest sustainable. I would expect the new wallet very soon.

Multipool is still on the same chain as the block crawler.  I've been checking it every few hours.

mullick,

Can you please give us a short update? Silence worries me.

I don't ask dates for fixes, just something like "we suspect something related to x" or "I need help from someone to do y" etc...

I am a SW engineer and starting to look at the CAP code (at my own slow but sustain pace). If there is an area you would like more eyeballs on, then please let me know.

Thanks.

It is in sync since morning

Has this been addressed yet? It looks like http://cap.coinmine.pl is still not in sync with the block crawler.

The problem will be fixed soon, and CAP will be back on track  

http://cap.coinmine.pl was again out of chain and night mining were lost... decision is yours guys - I will keep mining open but you should be aware about problems (I am sure you are )

Yea, 14 is the most connections I can get with Windows-7 HOME, using the 8-connects + the 8-addnodes trick... Guess another socket or thread is needed to extend that range for us.

I think two connections are the loopbacks... so it actually is 16 total, 8 and 8. But the loopback consumes 2 connections for RPC I guess.

If you mined coins on a fork... they don't actually exist... the fork rewards were not correct. They will show as a "?" in your wallet. However, tx's that were sent, will just be removed, unless the TX exists on both chains, or the sender resends them. (If you never get them, they return to the sender, by expiring in the senders wallet.)

If you go to your HELP menu, and go to the DEBUG WINDOW, then select the CONSOLE tab... Type the following...

repairwallet

If you mined coins on the wrong fork, it will show... Repaired 40 blocks, 400 coins removed/fixed. Or something like that.

However, if you are still not on the correct chain... and mining... you will have to do that again, after you get back on the correct chain.

I did as the guild in the first page, delete all file in Bottlecaps excluding wallet.dat & .conf file. Run again windows wallet 1.4.2, i could sync and lost all my coins?? I checked size of wallet.dat, nothing changed. How the way i can recover my coins?  hu hu

And that is why all my other machines are Linux.