09:27:51 GMT <fbred> Why can't I run BITFIELD read operatoins against my slaves? BITFIELD mykey GET u1 3212650845
09:27:59 GMT <fbred> (error) READONLY You can't write against a read only slave.
09:28:06 GMT <fbred> this makes no sense
09:52:08 GMT <Habbie> fbred, huh
10:00:15 GMT <fbred> Habbie, I am expecting to be able to read from my readslaves, but apparently, I can't with BITFIELD. Maybe because it can also by used to write data, I'm not sure. I can't see that it's documented, so I'm wondering if this is a bug or a fault on my part
10:00:37 GMT <Habbie> well from here it looks like a bug
10:00:39 GMT <Habbie> but i haven't tried it myself
10:00:44 GMT <Habbie> a normal GET on the field works?
10:07:54 GMT <fbred> Habbie, I think so, it's rather large, so it takes some time
10:09:23 GMT <fbred> if I redirect the output, I get a 1GB big file, so I guess GET works, but shouldn't it only be 512MB?
12:23:56 GMT <fbred> Habbie, so I looked at the source code and can confirm that this is a bug by design
12:24:07 GMT <Habbie> on purpose?
12:24:37 GMT <Habbie> oh let me guess, 'allowed on readonly' is decided purely based on the first word?
12:26:52 GMT <fbred> yes
12:26:58 GMT <Habbie> ah
12:27:07 GMT <Habbie> was wondering why i had no trouble with sorted sets
12:27:10 GMT <fbred> I will create a topic in google groups
12:27:11 GMT <Habbie> then guessed this
12:27:26 GMT <Habbie> thanks for this, this might actually have bitten me later on
12:28:00 GMT <fbred> the workaround is to make slaves accept writes
12:28:16 GMT <fbred> but that can create problems relating to key expires
12:28:58 GMT <fbred> maybe it would be best to create a read only bitfield command
12:29:12 GMT <Habbie> what's the key expire problem?
12:29:16 GMT <Habbie> i haven't looked at expiry at all yet
12:35:38 GMT <fbred> https://redis.io/topics/replication#read-only-slave
12:36:10 GMT <Habbie> there is GETBIT
12:36:12 GMT <Habbie> but it gives you single bits
12:36:26 GMT <Habbie> although you typed u1
12:37:29 GMT <Habbie> fbred, h
12:37:31 GMT <Habbie> fbred, ah
12:39:20 GMT <Habbie> fbred, do you know where the 'type' argument to BITFIELD is documented?
12:41:19 GMT <fbred> it's documented in the command, 'supported subcommands and integer types'
12:41:39 GMT <Habbie> oh i see it now
12:41:41 GMT <Habbie> thanks
12:42:00 GMT <Habbie> this appears to be the only part of redis that goes beyond 52/53 bits of precision for integers
12:47:17 GMT <fbred> I'm not sure about that, normal keys can be 64 bit signed integers
12:47:31 GMT <Habbie> aren't those treated as strings?
12:48:12 GMT <fbred> yes, but if you do for example incr on a string, it will be converted
12:48:29 GMT <Habbie> Note: this is a string operation because Redis does not have a dedicated integer type. The string stored at the key is interpreted as a base-10 64 bit signed integer to execute the operation.
12:48:33 GMT <Habbie> indeed
13:21:59 GMT <minus> did you get around to writing a redis module, Habbie?
13:22:05 GMT <Habbie> no :)
13:22:18 GMT <Habbie> it's nothing more than a vague idea for now
13:54:37 GMT <Mattias> can I set maxmemory policy per database or do I need to start a second redis instance for that?
19:53:16 GMT <mjb_> dear redis community, i wrote simple java app to export data from sql to redis as json becouse I need to export large amount data from postgresql, maybe it will be useful and helpful for someone of you, eg for test or development, https://github.com/mojeprojekty/sql2redis
20:48:04 GMT <dbkaplun> hey, is it possible to set an empty password?
20:48:35 GMT <minus> empty password means no password
20:48:56 GMT <dbkaplun> minus: is there docs for that?
20:49:12 GMT <dbkaplun> looking through the source and i can't find anything
20:49:14 GMT <minus> the default config, i guess
20:49:39 GMT <minus> and the fact that you can remove the password using `config set requirepass ""`
20:54:47 GMT <dbkaplun> is there really no documentation for this?
21:01:57 GMT <minus> sorry, i don't know if there is. i haven't come across any regarding config
21:02:05 GMT <minus> except the default config that is
21:02:57 GMT <badboy_> what documentation do you want?
21:03:57 GMT <badboy_> the empty password is actually a valid password
21:04:30 GMT <dbkaplun> so who's right?
21:04:45 GMT <badboy_> I tested it ;)
21:05:26 GMT <badboy_> 'config set requirepass ""' will indeed erase the password requirement
21:05:36 GMT <badboy_> but 'requirepass ""' in the config will set it to the empty string
21:06:43 GMT <dbkaplun> `-ERR wrong number of arguments for 'AUTH' command` when I try to send `AUTH     `
21:10:18 GMT <badboy_> yeah, well, you actually need to send the empty password^^
21:10:23 GMT <dbkaplun> so? who's right? lol
21:10:45 GMT <badboy_> everyone
21:10:48 GMT <dbkaplun> are there any open issues, etc about this? i feel this should be well-understood behavior
21:11:26 GMT <dbkaplun> badboy_: ok? how do i pass an empty password?
21:11:47 GMT <badboy_> redis-cli -a ""
21:11:53 GMT <badboy_> or in redis-cli: AUTH ""
21:11:59 GMT <badboy_> or in whatever client lib you use
21:12:04 GMT <dbkaplun> i'm using telnet
21:12:43 GMT <minus> not sure about you, but i'd consider this a bug
21:13:16 GMT <badboy_> *2<cr>$4<cr>auth<cr>$0<cr><cr>
21:13:18 GMT <badboy_> in telnet ;)
21:13:47 GMT <badboy_> minus: what? that the empty password is a valid password? my professor would disagree :D
21:14:20 GMT <badboy_> dbkaplun: the inline protocol can't handle empty strings, the binary-safe protocol does
21:14:30 GMT <minus> yes. especially that the behaviour is different between config and runtime
21:14:43 GMT <badboy_> that indeed
21:14:57 GMT <badboy_> but we do not have a "config delete password", so right now it's the only option
21:15:08 GMT <badboy_> (I'd be fine with disallowing the empty password in the config though)
21:15:15 GMT <badboy_> feel free to submit a PR ;)
21:15:21 GMT <minus> that's what i was suggesting
21:15:27 GMT <minus> empty password doesn't make sense anyway
21:15:44 GMT <dbkaplun> so wait. the empty password *is* a valid password?
21:16:21 GMT <badboy_> that's what I told you
21:16:27 GMT <dbkaplun> i know
21:16:42 GMT <badboy_> minus: again, my university professor would beg to differ :P
21:16:51 GMT <dbkaplun> all this code assumes empty password is no auth
21:17:03 GMT <badboy_> all of which code?
21:17:15 GMT <dbkaplun> nevermind
21:17:20 GMT <dbkaplun> is there documentation for this?
21:17:25 GMT <minus> badboy_: is he using an empty password…?
21:17:27 GMT <badboy_> as I told you: no
21:17:38 GMT <dbkaplun> is there somewhere i can look in the source?
21:17:46 GMT <minus> is his name stallman?
21:17:59 GMT <badboy_> minus: nah, just a stupid reference to my early lectures about algebra
21:18:09 GMT <dbkaplun> where is auth handling located in the source?
21:18:09 GMT <badboy_> one task was to give the number of passwords under some constraints
21:18:31 GMT <badboy_> and most students had the answer off by one because everyone automaitcally excluded the empty password
21:18:55 GMT <minus> some might call it a trick question
21:19:09 GMT <badboy_> check is here: https://github.com/antirez/redis/blob/unstable/src/server.c#L2548
21:19:12 GMT <minus> but those with experience call it knowing the edge cases
21:19:20 GMT <minus> damn edge cases!
21:19:40 GMT <badboy_> and config parser is here: https://github.com/antirez/redis/blob/unstable/src/config.c#L482-L487
21:21:00 GMT <badboy_> ah and there I guess :https://github.com/antirez/redis/blob/unstable/src/config.c#L805
21:21:07 GMT <badboy_> both places would need a >0 check, that's all
21:21:52 GMT <badboy_> (or ==0 → goto badfmt;)
21:22:46 GMT <dbkaplun> https://github.com/antirez/redis/issues/3676
21:22:54 GMT <dbkaplun> thanks
21:23:22 GMT <badboy_> you can document it yourself!
21:23:30 GMT <badboy_> I'm still on minus side: just disallow it
21:23:39 GMT <badboy_> breaking user code is highly unlikely