1 2013-01-17 00:00:06 <stealth222> to reduce the amount of roundtrip RPC
  2 2013-01-17 00:01:03 <stealth222> it's silly to use bitcoind's RPC to check whether an address has a correct base58checksum
  3 2013-01-17 00:02:03 <Acciaio> yes but validate address function answer well 90% of times
  4 2013-01-17 00:02:27 <stealth222> you still should validate all user inputs ALWAYS in a web application
  5 2013-01-17 00:02:36 <Acciaio> the real problem is walletpassphrase and the biggerone problem is sendfrom method
  6 2013-01-17 00:03:12 <stealth222> I concur that walletpassphrase sucks for this type of usage
  7 2013-01-17 00:03:37 <stealth222> probably better to not bother encrypting the wallet and instead placing stricter access controls via some RPC proxy
  8 2013-01-17 00:04:10 <BlueMatt> that and do a hotwallet vs savings wallet distinction
  9 2013-01-17 00:04:10 <stealth222> if someone manages to hack into your bitcoind server, chances are your bitcoins are already gone anyhow
 10 2013-01-17 00:04:14 <sipa> or encypting the file, but unlocking it once for a very long time
 11 2013-01-17 00:04:35 <sipa> so at least people can't grab the wallet if they just get access to the wallet.dat file
 12 2013-01-17 00:05:10 <stealth222> walletpassphrase is ok for interactive usage of the wallet. for RPC, per-connection stateless secret stuff is way better
 13 2013-01-17 00:06:45 <stealth222> and as BlueMatt says, keep a hotwallet with only sufficient funds to remain solvent - and transfer into it manually from a more secure wallet when funds are running low
 14 2013-01-17 00:11:22 <Acciaio> yes I use the bluematt solution a little hotspot wallet ... but for sure I have to study something about security
 15 2013-01-17 00:11:54 <Acciaio> however now there is almost nothing to steal
 16 2013-01-17 00:12:10 <stealth222> so then don't bother encrypting the wallet
 17 2013-01-17 00:12:19 <stealth222> encrypt it only to make backups
 18 2013-01-17 00:12:24 <BlueMatt> dont ever underestimate how little effort a bored hacker will go to to steal "almost nothing" :)
 19 2013-01-17 00:12:42 <BlueMatt> s/little/much/
 20 2013-01-17 00:13:33 <stealth222> also, even if the wallet is encrypted, if a hacker breaks into your web server, unless you secure the RPC with some access controls, they can still steal everything
 21 2013-01-17 00:13:50 <Acciaio> yes stealth but I have encrypted the wallet only to find a solution to sendfrom proble
 22 2013-01-17 00:14:00 <Acciaio> I was crazy today to make it work
 23 2013-01-17 00:14:04 <stealth222> for instance, place limits on volume and number of transactions - and set alerts when strange things happen
 24 2013-01-17 00:14:09 <stealth222> and audit EVERY CALL
 25 2013-01-17 00:14:55 <stealth222> and back up the audit to a remote machine regularly
 26 2013-01-17 00:15:26 <stealth222> that alone will do FAR more than walletpassphrase :p
 27 2013-01-17 00:16:39 <Acciaio> wallet passphrase suck because I have to store the passphrase somewhere
 28 2013-01-17 00:17:00 <Acciaio> so I can read it from php
 29 2013-01-17 00:17:19 <BlueMatt> or run a script on the bitcoind machine that just unlocks the wallet
 30 2013-01-17 00:17:30 <sipa> neither does make sense
 31 2013-01-17 00:17:53 <sipa> if an attacker can access the password as easily as the wallet.dat file, encryption is pointless
 32 2013-01-17 00:18:14 <BlueMatt> heh dur yes
 33 2013-01-17 00:18:29 <sipa> what does make sense is encrypting it, but only unlock it manually (after startup of your system), with a very long timeout
 34 2013-01-17 00:19:29 <Acciaio> sipa, if I unlock it manually it will stay unlocked also for rpc calls????
 35 2013-01-17 00:19:30 <Acciaio> ...
 36 2013-01-17 00:19:33 <lianj> does restarting bitcoind clear its orphan txpool pool?
 37 2013-01-17 00:19:36 <sipa> Acciaio: yes
 38 2013-01-17 00:19:45 <sipa> lianj: yes
 39 2013-01-17 00:19:48 <Acciaio> I have some confusion about this
 40 2013-01-17 00:19:49 <lianj> sipa: merci
 41 2013-01-17 00:20:14 <sipa> Acciaio: the point of walletpassphrase is to prevent the secret key for needing to be in memory longer than necessary
 42 2013-01-17 00:20:21 <sipa> it is not access control
 43 2013-01-17 00:20:29 <sipa> (though it can be used as a weak form of that)
 44 2013-01-17 00:21:33 <sipa> Acciaio: the key just needs to be available when sending the coins - it doesn't matter how bitcoind knows it
 45 2013-01-17 00:22:33 <sipa> Acciaio: and "manually" would still be an RPC call, but perhaps one made by bitcoind itself
 46 2013-01-17 00:23:17 <sipa> or even better, with a script that asks you for the passphrase instead of you needing to type in on the command line (where it might accidentally end up in .bash_history)
 47 2013-01-17 00:43:38 <Acciaio> thanks to all I have to go now!
 48 2013-01-17 00:43:47 <Acciaio> see you soon bye
 49 2013-01-17 02:30:09 <stealth222> are any of the bitcoin-qt people around?
 50 2013-01-17 02:34:16 <wiepe> what kind of person in his right frame of mind would talk to this gmaxwell character?
 51 2013-01-17 02:36:20 <eckey> touchy
 52 2013-01-17 02:49:19 <BlueMatt> ass hole's been bugging gmaxwell for a while...
 53 2013-01-17 02:49:36 <BlueMatt> well, and this channel
 54 2013-01-17 02:50:04 <eckey> and who is jgarzik?
 55 2013-01-17 02:50:31 <BlueMatt> one of the bitcoin core developers?
 56 2013-01-17 02:52:59 <eckey> but not andreson or hearn?
 57 2013-01-17 02:54:27 <BlueMatt> you have a weird list...
 58 2013-01-17 02:54:54 <BlueMatt> andresen leads satoshi client, but hearn does bitcoinj
 59 2013-01-17 02:54:56 <eckey> I've been away for two years.  Please explain...
 60 2013-01-17 02:55:24 <BlueMatt> there are waaaay more than like 2 developers on bitcoin, and more than one client
 61 2013-01-17 02:55:29 <BlueMatt> each with multiple developers
 62 2013-01-17 02:57:13 <eckey> and each of those developers is empowered to kick others off this list?  Just trying to understand...
 63 2013-01-17 02:57:35 <BlueMatt> its an irc channel, people kick spammers...calm down
 64 2013-01-17 03:01:09 <eckey> I'm sitting here with a bottle of Old Bushmills and a shot glass--I am very calm.  Just trying to understand how one nobody can kick another nobody.
 65 2013-01-17 03:01:35 <BlueMatt> ehhh...not at all, one very much somebody kicked a very, very regular spammer
 66 2013-01-17 03:02:10 <BlueMatt> anyway, I hate discussions sparked by spammers/trolls, so Im gonna stop now
 67 2013-01-17 03:02:50 <eckey> weipe's comment was "what kind of person in his right frame of mind would talk to this gmaxwell character?" and he got kicked.  Maybe you should suggest jgarzik calm down...
 68 2013-01-17 03:03:00 <BlueMatt> lol
 69 2013-01-17 03:03:01 <BlueMatt> ok
 70 2013-01-17 03:03:09 <eckey> spammer?  what is he selling?
 71 2013-01-17 03:03:25 <Luke-Jr> eckey: he comes in here and harrasses gmaxwell every other day
 72 2013-01-17 03:03:31 <phantomcircuit> s/other//
 73 2013-01-17 03:03:40 <phantomcircuit> the guy really has no life
 74 2013-01-17 03:03:57 <eckey> Hi Luke, I remember you from years ago.  Who is this gmaxwell?
 75 2013-01-17 03:04:17 <Luke-Jr> eckey: bitcoind developer
 76 2013-01-17 03:04:22 <Luke-Jr> eckey: Gregory Maxwell
 77 2013-01-17 03:05:24 <eckey> ok.  just trying to get a handle on the politics of the #bitcoin channels
 78 2013-01-17 03:05:32 <Luke-Jr> meh, no real politics
 79 2013-01-17 03:05:42 <Luke-Jr> obvious trolls get banned, that's pretty much it
 80 2013-01-17 03:07:43 <eckey> I'm developing wallet software for a new web site, FWIW
 81 2013-01-17 03:08:02 <eckey> ECKey, etc.
 82 2013-01-17 03:24:25 <Luke-Jr> jgarzik: perhaps /ban mughat3!*@*!##fix_your_connection
 83 2013-01-17 05:00:41 <jgarzik> channel ban list is full?  sigh
 84 2013-01-17 05:03:12 <petertodd> jgarzik: btw, I'm working on doing that cython port of pynode I mentioned before
 85 2013-01-17 05:03:27 <jgarzik> petertodd: cool
 86 2013-01-17 05:04:03 <petertodd> I'm calling it python-bitcoin, and I ripped out all the node-specific stuff to make it just a library
 87 2013-01-17 05:04:24 <jgarzik> petertodd: that's pretty much already the goal behind the entire bitcoin/ directory ;p
 88 2013-01-17 05:04:53 <petertodd> heh, yeah I figured as much, it's decently well organized
 89 2013-01-17 05:05:53 <petertodd> basically I deleted everything that wasn't in bitcoin/ :P
 90 2013-01-17 05:08:00 <jgarzik> petertodd: ideal would have been a pull request moving the node stuff, then adding __init__.py to bitcoin/ and making it a proper package
 91 2013-01-17 05:08:55 <petertodd> jgarzik: yeah, I figured it's all pretty invasive, and it looked like it hadn't been worked on for awhile, so I'd see what makes sense before doing the pull and stuff like that
 92 2013-01-17 05:09:40 <jgarzik> petertodd: it's actively used... not too many bugs in the core
 93 2013-01-17 05:10:44 <petertodd> heh, I found a few by compiling it that I need to do pull reqs for :P what is it getting used for btw?
 94 2013-01-17 05:11:53 <jgarzik> petertodd: Odd Jobs </goldfinger>
 95 2013-01-17 05:12:05 <petertodd> lol
 96 2013-01-17 05:12:40 <jgarzik> petertodd: several people have used it on their websites as a secondary monitoring node, or for things that are easily scriptable in python but not in bitcoind -- notably event notifications like block/tx async notifications
 97 2013-01-17 05:13:38 <petertodd> cool, yeah, I mainly want a library with a pythonic interface to script's and transactions
 98 2013-01-17 05:14:31 <petertodd> and with cython I suspect it can be just as fast as libbitcoin
 99 2013-01-17 05:15:40 <jgarzik> should be fast enough for just about anything besides full-chain-verification
100 2013-01-17 05:16:19 <jgarzik> anyway, well past time for sleep.  send pull requests, even major ones!  *poof*
101 2013-01-17 05:16:26 <petertodd> ha, will do
102 2013-01-17 06:59:09 <mariusursache> hello. I'm using bitcoin-qt for mac to generate some test coints (on testnet), however since 2 days ago I have 2k BTC as 'immature'. can I do something to move them to balance?
103 2013-01-17 07:15:46 <weex> mariusursache: has it been 100 or 120 blocks since they were generated?
104 2013-01-17 07:16:14 <mariusursache> where can I check how many blocks were?
105 2013-01-17 07:16:41 <gribble> 216880
106 2013-01-17 07:16:41 <SomeoneWeird> ;;bc,blocks
107 2013-01-17 07:17:26 <mariusursache> ;;bc,blocks
108 2013-01-17 07:17:27 <gribble> 216880
109 2013-01-17 07:17:31 <mariusursache> ;;help
110 2013-01-17 07:17:32 <gribble> The bot responds when you start a line with the ! character. A good starting point for exploring the bot is the !facts command. You can also visit the bot's website for a list of help topics and documentation: http://gribble.sourceforge.net/
111 2013-01-17 07:19:13 <weex> mariusursache: does the list of transactions in bitcoin-qt show how many confirmations since those coins were generated?
112 2013-01-17 07:19:36 <mariusursache> weex: number of transactions: 43
113 2013-01-17 07:19:45 <mariusursache> is this the confirmation?
114 2013-01-17 07:20:32 <weex> nope, i thought it showed confirmations on the left of each transaction
115 2013-01-17 07:20:37 <weex> but perhaps for generation it doesnt
116 2013-01-17 07:20:41 <mariusursache> I think I found it. in the transactions tab, on some it says 91/92 confirmations. when those get to 100 they are confirmed?
117 2013-01-17 07:20:52 <weex> or 120
118 2013-01-17 07:21:14 <mariusursache> cool, thanks. so on testnet is that slow. how fast is it on realnet?
119 2013-01-17 07:21:16 <weex> but might as well check in an couple hours
120 2013-01-17 07:21:31 <weex> 1 confirmation every 10 mins on avg
121 2013-01-17 07:21:47 <weex> so 144 per day
122 2013-01-17 07:21:52 <mariusursache> that's 10 min on real net or on testnet?
123 2013-01-17 07:21:56 <SomeoneWeird> both
124 2013-01-17 07:21:57 <weex> real
125 2013-01-17 07:21:59 <SomeoneWeird> it's average
126 2013-01-17 07:22:19 <mariusursache> thank you
127 2013-01-17 07:22:26 <weex> i didn't know if testnet had enough hashpower to stick to 10 minx
128 2013-01-17 07:22:28 <weex> mins*
129 2013-01-17 07:22:41 <SomeoneWeird> pretty sure it's still 10 on the testnet
130 2013-01-17 08:09:23 <muhoo> are there compiled jars out there of bitcoinj 0.6.x anywhere?
131 2013-01-17 08:09:52 <muhoo> i found jars of 0.7, but it conflicts with a bunch of stuff (guano, etc)
132 2013-01-17 09:43:09 <mariusursache> I think on testnet the speed is 1 confirmation at each 30 minutes. I had 92 earlier, it's 96 now. so 4 in 2h20min
133 2013-01-17 09:51:06 <sipa> mariusursache: the rule is 1 block per 10 minutes
134 2013-01-17 09:51:12 <sipa> mariusursache: but there must be someone mining
135 2013-01-17 09:51:30 <sipa> if nobody is mining on testnet (or at very variable speed), confirmations are slower of course
136 2013-01-17 09:52:03 <mariusursache> sipa: mining with some gpu/fpga or should work with regular computer?
137 2013-01-17 09:52:22 <sipa> yes
138 2013-01-17 09:52:39 <sipa> difficulty is far lower on testnet
139 2013-01-17 09:53:28 <mariusursache> so I can start an ubuntu vm and do some mining?
140 2013-01-17 09:54:21 <sipa> you can even mine just with CPU on testnet; start the client with -gen
141 2013-01-17 10:27:14 <t7> is anyone on the bitcoin dev team being hired to work on bitcoin?
142 2013-01-17 10:31:08 <sipa> t7: gavin is paid by the bitcoin foundation
143 2013-01-17 10:56:50 <dparrish> ;;ticker
144 2013-01-17 10:56:50 <gribble> BTCUSD ticker | Best bid: 14.82186, Best ask: 14.84100, Bid-ask spread: 0.01914, Last trade: 14.84100, 24 hour volume: 43600.45820899, 24 hour low: 14.42512, 24 hour high: 14.92000, 24 hour vwap: 14.67562
145 2013-01-17 12:48:57 <Yrouel> hi
146 2013-01-17 12:50:01 <Yrouel> I'm using bitcoin-qt on mac and I'd like to have bitcoind as client in background so when I open the gui I can find it alread synced with the network. Is that possible?
147 2013-01-17 12:51:12 <luke-jr_> Yrouel: no
148 2013-01-17 12:51:32 <Yrouel> I don't need to use the gui too often but if I don't leave it opened next time I do it takes forever to sync so with a deamon inbackground (possibly integrated with launchd) it would sync silently and the app wouldn't require so much time then
149 2013-01-17 12:51:42 <luke-jr_> well, sortof
150 2013-01-17 12:51:49 <luke-jr_> you could shutdown bitcoind before starting Bitcoin-Qt
151 2013-01-17 12:52:04 <Yrouel> soo
152 2013-01-17 12:52:05 <Yrouel> basically
153 2013-01-17 12:52:16 <luke-jr_> that'd keep the blockchain sync'd, but you still need to exit/load it
154 2013-01-17 12:52:26 <Yrouel> even if bitoind is bundled with bitcoind-qt the integration between the two is non existant?
155 2013-01-17 12:52:33 <luke-jr_> correct
156 2013-01-17 12:53:26 <Yrouel> well could you add that feature? it would at least "mask" the problem of slow sync
157 2013-01-17 12:54:04 <luke-jr_> ???
158 2013-01-17 12:54:17 <luke-jr_> there's no reason not to just leave Bitcoin-Qt running 24/7
159 2013-01-17 12:54:44 <Yrouel> well but that's a job for a daemon
160 2013-01-17 12:54:52 <Yrouel> would be a proper implementation of that logic
161 2013-01-17 12:55:03 <luke-jr_> perhaps, but that's not how Bitcoin-Qt is designed
162 2013-01-17 12:55:18 <luke-jr_> patches welcome, I'm pretty sure
163 2013-01-17 12:55:35 <Yrouel> is that another gui tat acts more as a frontend for bitcoind?
164 2013-01-17 12:55:51 <Yrouel> is there*
165 2013-01-17 12:56:10 <luke-jr_> there was Spesmilo, but everyone stopped maintaining it
166 2013-01-17 12:58:47 <Yrouel> luke-jr_ is thanks for the info
167 2013-01-17 12:59:00 <Yrouel> is there a schedule plan for next release?
168 2013-01-17 12:59:15 <Yrouel> 0.8 if I'm right
169 2013-01-17 12:59:42 <Yrouel> that it should at least implement something new to speed up the database or something like that right?
170 2013-01-17 12:59:55 <luke-jr_> no;yes
171 2013-01-17 13:00:18 <Joric> maybe someone knows, i have an OCZ agility 3 ssd drive, does it use ATA password in the BIOS as a base for its encryption keys? does it even support this kind of encryption?
172 2013-01-17 13:00:47 <Diablo-D3> no. probably not.
173 2013-01-17 13:00:52 <Yrouel> luke-jr_ historically what's the avg time between releases?
174 2013-01-17 13:01:33 <Joric> Diablo-D3, just a shot in a dark? )
175 2013-01-17 13:02:01 <Diablo-D3> Joric: if it has built in encryption, it came with windows utilities to do it
176 2013-01-17 13:02:36 <Diablo-D3> Joric: no consumer drive has it afaik
177 2013-01-17 13:02:47 <Diablo-D3> very few enterprise drives have it, theres no call for it
178 2013-01-17 13:02:55 <Diablo-D3> its easier to just do it in a dedicated hardware controller
179 2013-01-17 13:03:12 <sipa> Yrouel: you're very welcome to try test builds for 0.8, see https://bitcointalk.org/index.php?topic=129861.0
180 2013-01-17 13:04:21 <Joric> heard truecrypt drastically affects ssd speed + not secure at all because it can't overwrite old keys due to wear leveling
181 2013-01-17 13:05:11 <Yrouel> sipa uhm how "official" are those?
182 2013-01-17 13:05:34 <Luke-Jr> Yrouel: there's no such thing as official for Bitcoin
183 2013-01-17 13:05:39 <Joric> * same about compromised wallet keys ) if you're using ssd
184 2013-01-17 13:06:00 <Luke-Jr> Yrouel: sipa/Pieter is the main developer of the major changes in 0.8
185 2013-01-17 13:06:11 <Yrouel> Luke-Jr well bitcoin.org is the "official" site for example
186 2013-01-17 13:06:22 <Yrouel> anyway I don't see prebuilt mac versions so :(
187 2013-01-17 13:06:30 <Luke-Jr> Mac is always a pain to deal with
188 2013-01-17 13:06:48 <Yrouel> why?
189 2013-01-17 13:06:57 <Yrouel> using qt shouldn't be too hard
190 2013-01-17 13:07:09 <Luke-Jr> Yrouel: you know how to setup a cross-compiler? :P
191 2013-01-17 13:07:22 <Luke-Jr> I've put hours into it and I still don't have it working
192 2013-01-17 13:07:24 <Yrouel> oh cross compiling is a pita
193 2013-01-17 13:07:28 <Luke-Jr> only for Mac
194 2013-01-17 13:08:02 <Scrat> Joric: ATA password support for SSDs (and M/Bs) is a sad joke
195 2013-01-17 13:08:04 <Joric> Diablo-D3, before you i was practically sure that all ssd's support encryption you only have to change ata password in the bios :(
196 2013-01-17 13:08:12 <Scrat> and vendors won't even tell you what implementation they use
197 2013-01-17 13:08:57 <Yrouel> Luke-Jr I guess the pita is the crosscompilation part, compiling the code directly on a mac I thing would be much less painful
198 2013-01-17 13:09:00 <Joric> Scrat, how it works then? i have ocz agility 3
199 2013-01-17 13:09:10 <Luke-Jr> Yrouel: then we need a Mac etc :p
200 2013-01-17 13:09:24 <Yrouel> yeah it's the little detail gh
201 2013-01-17 13:09:32 <Scrat> can your BIOS set ata password? the only way to try it is to set it and then plug it on another pc
202 2013-01-17 13:09:36 <sipa> Yrouel: I maintain those test builds; they're not official as in not verified by several developers (as normal releases do), and it's build with patches that aren't accepted in mainline; otherwise, they are built in exactly the same way, and are gpg signed by me (and my GPG key is on http://bitcoin.org/pieterwuille.asc)
203 2013-01-17 13:10:11 <Yrouel> sipa ok but there isn't a mac buils so thanks anyway :)
204 2013-01-17 13:10:14 <Luke-Jr> Yrouel: 90% of us use Linux, and 1 uses Windows ;)
205 2013-01-17 13:10:31 <Luke-Jr> Yrouel: "official" Windows binaries are built on Linux in a deterministic way (so the SHA256 of them is identical)
206 2013-01-17 13:10:35 <sipa> Yrouel: oh, sorry, I don't do any OSX stuff - you can build them yourself if you need to, though
207 2013-01-17 13:10:37 <Joric> Scrat, got ya, i'll try, yes i have separate hdd password not sure it arrived with ssd or always been there
208 2013-01-17 13:11:13 <Yrouel> sipa yep I know I still haven't setup a proper qt buildroot
209 2013-01-17 13:11:43 <Scrat> I would not trust any closed implementation with encrypting my data. for all you know they are storing the password on the disk
210 2013-01-17 13:11:52 <Luke-Jr> Yrouel: but you just said it was easy :o
211 2013-01-17 13:12:22 <Yrouel> Luke-Jr it is, I just haven't the patience to download the framework and install it
212 2013-01-17 13:12:31 <Yrouel> because 'till now the only reason would have been a single app
213 2013-01-17 13:12:34 <Scrat> truecrypt on ssd might have some overhead and might wear your disk out sooner but I don't see another solution
214 2013-01-17 13:12:36 <Yrouel> now there's two
215 2013-01-17 13:12:43 <Yrouel> and it might be worth it
216 2013-01-17 13:12:45 <Scrat> eCrypfts is great if you can use it (linux)
217 2013-01-17 13:12:50 <Joric> Scrat, oh those closed source encryption schemes
218 2013-01-17 13:13:26 <Joric> "Since key length and key structure vary and since the encryption engine does not use any mathematical algorithms, reverse engineering is impossible and guessing is not an option."
219 2013-01-17 13:13:44 <Joric> http://www.engr.mun.ca/~howard/crypto2012/crypto_snake_oil.pdf
220 2013-01-17 13:13:44 <Luke-Jr> XD
221 2013-01-17 13:14:00 <Luke-Jr> Joric: dead link ;)
222 2013-01-17 13:14:05 <Joric> dam
223 2013-01-17 13:14:11 <Joric> google crypto snake oil
224 2013-01-17 13:21:15 <Yrouel> Luke-Jr qt 5 is ok or I need a previous version?
225 2013-01-17 13:21:43 <sipa> Yrouel: 4.8
226 2013-01-17 13:27:11 <Diablo-D3> [09:08:29] <Joric> Diablo-D3, before you i was practically sure that all ssd's support encryption you only have to change ata password in the bios :(
227 2013-01-17 13:27:21 <Diablo-D3> dude, my bios doesnt even have an "ata password"
228 2013-01-17 13:27:37 <Scrat> most bioses dont
229 2013-01-17 13:27:40 <Diablo-D3> [09:04:46] <Joric> heard truecrypt drastically affects ssd speed + not secure at all because it can't overwrite old keys due to wear leveling
230 2013-01-17 13:27:43 <Diablo-D3> not entirely true
231 2013-01-17 13:27:49 <Diablo-D3> truecrypt writes encrypted data to the drive
232 2013-01-17 13:28:07 <Diablo-D3> yes, it leaves old encrypted data behind due to wear leveling, but its hard for someone to get that data
233 2013-01-17 13:28:16 <Diablo-D3> you're looking at a government agency trying to recover it at that point
234 2013-01-17 13:28:18 <Diablo-D3> AAAAAAAAND
235 2013-01-17 13:28:20 <Diablo-D3> its still encrypted.
236 2013-01-17 13:28:47 <Diablo-D3> and it doesnt effect ssd speed, it effects cpu bound write throughput
237 2013-01-17 13:28:52 <Diablo-D3> ie, your cpu might not be fast enough.
238 2013-01-17 13:29:09 <Joric> those drives should have an emergency button "flash it all at once"
239 2013-01-17 13:29:21 <Diablo-D3> most drives have secure destroy
240 2013-01-17 13:29:25 <Scrat> true that, a FS cannot tell an SSD to overwrite data because it will map to different LBAs
241 2013-01-17 13:29:57 <Diablo-D3> btw, real harddrives use wear leveling too
242 2013-01-17 13:30:04 <Diablo-D3> Im not sure why people suddenly realized wear leveling existed
243 2013-01-17 13:30:25 <Scrat> or rather same LBA will result to different physical chip writes
244 2013-01-17 13:30:33 <Diablo-D3> Scrat: yes, that
245 2013-01-17 13:30:45 <Diablo-D3> the point I was making is your garbage is still encrypted
246 2013-01-17 13:30:47 <Joric> well not 'data', i was speaking about old volume header with compromised keys i guess it can be found by signature
247 2013-01-17 13:30:47 <Scrat> Diablo-D3 pretty sure hdds do that only if there's a problem with a block
248 2013-01-17 13:30:53 <Diablo-D3> Scrat: nope
249 2013-01-17 13:30:53 <Scrat> aka block reallocation
250 2013-01-17 13:31:04 <erska> encryption may affect ssd speed, some drives compress the data on the fly before writing it, and encrypted data should not compress at all
251 2013-01-17 13:31:05 <Diablo-D3> they wear level as well
252 2013-01-17 13:31:07 <Scrat> and keep a small map of reallocations
253 2013-01-17 13:31:26 <Diablo-D3> but its within a local area
254 2013-01-17 13:31:31 <Yrouel> sipa I see some merged pull requests about qt5 is still not supported tho?
255 2013-01-17 13:31:51 <Diablo-D3> erska: drives that compress should be considered scams
256 2013-01-17 13:32:08 <Diablo-D3> virtually all the data people have _is already compressed_
257 2013-01-17 13:33:00 <Joric> anyway, using truecrypt on ssd drives is questionable
258 2013-01-17 13:33:10 <erska> video files, images, archives yeah, they are compressed
259 2013-01-17 13:33:16 <Scrat> oh snap
260 2013-01-17 13:33:17 <erska> but those are not things SSDs are usually used for
261 2013-01-17 13:33:30 <Scrat> apparently with kernel 3.1+ dmcrypt can do TRIM passthrough
262 2013-01-17 13:33:31 <erska> more often people install things that need to load fast
263 2013-01-17 13:33:33 <erska> OS, apps, games
264 2013-01-17 13:33:38 <erska> and those usually do compress
265 2013-01-17 13:33:42 <Diablo-D3> Joric: its not questionable for the reasons I just said
266 2013-01-17 13:34:03 <erska> and when they do compress, it saves some of the wear from flash, since the drive does not need to write as much
267 2013-01-17 13:34:06 <Diablo-D3> erska: games are already compressed
268 2013-01-17 13:34:17 <Diablo-D3> and drives simply dont have the hardware for real compression
269 2013-01-17 13:34:19 <erska> not always
270 2013-01-17 13:34:31 <Diablo-D3> erska: game assets are already compressed.
271 2013-01-17 13:34:37 <Scrat> installed games? not really. some of them use compressed textures
272 2013-01-17 13:35:25 <Diablo-D3> Scrat: even if its uncompressed, a lot of games use zips as the container
273 2013-01-17 13:36:06 <erska> I have a typical Steam installation here with about 20 games, uncompressed size is 75GB, NTFS compressed size is 56,7GB
274 2013-01-17 13:36:12 <erska> and NTFS does not do very heavy compression
275 2013-01-17 13:36:26 <Joric> what are you talking about - games, videos ) most ppl store databases and corporate source code
276 2013-01-17 13:36:46 <erska> so even writing that to SSD with similar compression ratio would save about 20GB of writes
277 2013-01-17 13:36:49 <Diablo-D3> lol ntfs compression
278 2013-01-17 13:36:52 <Diablo-D3> lolololololol
279 2013-01-17 13:37:09 <Diablo-D3> man, whats with noobs lately
280 2013-01-17 13:37:11 <erska> compression type is not the point here
281 2013-01-17 13:37:21 <erska> but the point that installed games usually do compress
282 2013-01-17 13:37:48 <Diablo-D3> and saving 20gb of written once data does... nothing.
283 2013-01-17 13:38:48 <Joric> i wonder why ssd compression wasn't used for the marketing yet :)
284 2013-01-17 13:39:19 <Scrat> oh it is
285 2013-01-17 13:39:29 <Scrat> it brings their write amplication down
286 2013-01-17 13:40:01 <Scrat> this is the kind of marketing for an anandtech reader tho, lol
287 2013-01-17 13:41:51 <Diablo-D3> yeah, I wonder if it can be turned off
288 2013-01-17 13:41:56 <Diablo-D3> I wont buy a drive with it
289 2013-01-17 13:44:07 <Diablo-D3> seriously, why the fuck would I want a device with a tiny little microcontroller in it encrypting
290 2013-01-17 13:44:11 <Diablo-D3> that makes no goddamned sense
291 2013-01-17 13:45:32 <erska> then you don't need to wipe every flash block when doing a secure erase
292 2013-01-17 13:45:37 <erska> they just wipe the encryption key
293 2013-01-17 13:46:41 <Joric> meh, don't be ridiculous
294 2013-01-17 13:46:47 <Diablo-D3> lol
295 2013-01-17 13:47:02 <Diablo-D3> also, I meant compressing
296 2013-01-17 13:47:07 <Diablo-D3> same goes for encrypting
297 2013-01-17 13:47:18 <Diablo-D3> theres no reason to have a drive do it
298 2013-01-17 13:47:26 <Diablo-D3> theres no way in hell it can be fast enough
299 2013-01-17 13:48:16 <erska> simple, it reduces flash write cycles, thus extending life of the drive
300 2013-01-17 13:48:27 <erska> unless all you write is incompressible data
301 2013-01-17 13:48:44 <Diablo-D3> most of the data on people's drive is a) written once b) already compressed
302 2013-01-17 13:48:54 <Diablo-D3> so why would you _slow the drive down_ being dumb?
303 2013-01-17 13:50:04 <sipa> if you add a hardware encryption chip that can sustain the bandwidth of the drive itself, there will not be any measurable slowdown
304 2013-01-17 13:50:24 <sipa> as in: only the extra latency of that encryption chip, but that's likely microseconds
305 2013-01-17 13:50:31 <Joric> stop that ssd flame already i just asked if it somehow utilizes bios password or not :)
306 2013-01-17 13:51:07 <Diablo-D3> sipa: yes, but thats a $500 raid controller
307 2013-01-17 13:51:30 <Joric> apparently i just have to try it myself - set a password, then try this ssd on another machine
308 2013-01-17 13:51:36 <Scrat> bleh its all about write amplfication, and decent compression isnt expensive to put on an arm chip nowadays
309 2013-01-17 13:52:39 <erska> you don't need a $500 raid controller, sandforce-based drives do compression and encryption in hardware, at wire speeds
310 2013-01-17 13:53:11 <erska> if you don't have enough flash chips in the drive to sustain wire speeds with uncompressed data, compressing actually speeds the drive up
311 2013-01-17 13:54:58 <erska> and that is why I said in the start that encryption may affect ssd speeds, since any of the data going to the ssd won't compress any further
312 2013-01-17 13:55:01 <Joric> sandforce logo looks like stark industries hehe http://en.wikipedia.org/wiki/SandForce
313 2013-01-17 13:56:53 <Joric> "and "AES encryption" [4] which works in the background and is completely automatic. It is linked to the BIOS password and encrypts the user data at the full speed of the data as it passes through the controller." -- wiki
314 2013-01-17 14:01:16 <kjj> you always always always compress first, and then encrypt second
315 2013-01-17 14:11:35 <Joric> i wonder will blkindex.dat be splitted when it reaches 2gb?
316 2013-01-17 14:11:50 <Joric> it's already over 1gb
317 2013-01-17 14:14:26 <sipa> Joric: i don't think so
318 2013-01-17 14:14:34 <sipa> Joric: i doubt BDB automatically splits database files
319 2013-01-17 14:14:55 <Joric> so, we have another possibly critical issue? )
320 2013-01-17 14:15:06 <sipa> 0.8 won't use blkindex.dat anymore :p
321 2013-01-17 14:16:32 <SomeoneWeird> :o
322 2013-01-17 14:18:33 <sipa> and the LevelDB databases to replace it use a directory per database, with many files that each are at most a few MB
323 2013-01-17 15:23:33 <sipa> seems by DNS seed is serving over 5 requests/s now... last time i checked (which may be several months ago) that was only 1-2
324 2013-01-17 15:23:58 <sipa> *my
325 2013-01-17 15:24:10 <kjj> that seems like a good metric to keep track of
326 2013-01-17 15:25:14 <kjj> do you just pull that out of your peers list, or do you update it manually?
327 2013-01-17 15:26:34 <sipa> kjj: i have written a custom p2p crawler + dns seeder
328 2013-01-17 15:26:48 <sipa> that keeps statistics about reachability of IPs and such
329 2013-01-17 15:26:57 <sipa> it badly needs a rewrite though
330 2013-01-17 15:28:43 <sipa> or wait... are you referring to the request speed?
331 2013-01-17 15:29:13 <sipa> this is typical output:
332 2013-01-17 15:29:14 <sipa> [13-01-17 17:29:11] 2681/455211 available (451334 tried in 72958s, 3853 new, 24 active), 17067 banned; 7637 DNS requests, 144 db queries
333 2013-01-17 15:29:44 <sipa> the last two numbers are since restart, which was less than half an hour ago
334 2013-01-17 16:06:11 <sipa> gavinandresen: after some investigation, it seems that's openssl's ecdsa verify performance is around 2 times slower on 32-bit
335 2013-01-17 16:06:30 <sipa> thanks to stealth222 for running on a few systems too :)