**** BEGIN LOGGING AT Tue Feb 02 02:59:58 2016
Feb 02 03:43:50 <brolin_empey> The default configuration of Microsoft Excel for Windows version 14 (Excel 2010) appears to lack a basic feature that Quattro Pro for Windows (QPW) version 17 (from year 2014) has in the default configuration:  QPW draws a blue triangle in the bottom left corner of a cell when the value of the cell is the result of a calculation, including a calculation contained entirely within the cell.  I know that Microsoft Office 2016 is the current release but 
Feb 02 03:43:50 <brolin_empey> Microsoft Office 2010 is the last release for Windows XP.  Corel WordPerfect Office 17 from year 2014 still supports Windows XP.  I personally use only Windows 7 and higher but many persons, especially in a corporate or other managed environment, are still stuck on Windows XP (or even earlier).
Feb 02 03:49:07 <brolin_empey> Is Excel 2016 still missing this feature of QPW 17?
Feb 02 03:53:05 <brolin_empey> Is there any spreadsheet software that, when selecting or editing a cell, can recursively highlight all of the cells used to calculate the value in the selected cell instead of highlighting only the cells directly referenced by the highlighted cell?
Feb 02 03:57:04 <brolin_empey> Is this a feature that is too useful to be implemented in practice, like the ability to subscribe to my bank account as an RSS or Atom news feed?  Why can I subscribe to the blog of someone who does not even know of me as a machine-readable news feed but I cannot subscribe to something private but that is as common as a bank account?
Feb 02 03:57:45 <robbiethe1st> Yes on both
Feb 02 04:00:54 <brolin_empey> I asked 3 questions, not only 2 questions.
Feb 02 04:56:16 <brolin_empey> I wanted to see if QPW version 10 from year 2001 (from Corel WordPerfect Office 2002) has this same feature as QPW 17 but QPW 10 always crashes with STATUS_HEAP_CORRUPTION ((NTSTATUS)0xC0000374L) when I try to open the document I created with and saved from QPW 17.
Feb 02 05:22:24 <brolin_empey> Neither Gnumeric nor LibreOffice Calc seems to have this feature of QPW.
Feb 02 05:29:30 <brolin_empey> KotCzarny: Can you show me where the specific model of mSATA↔PATA converter you use is offered for sale?
Feb 02 05:30:31 <brolin_empey> KotCzarny: I can try to find it by myself but I will not know if it is the same model as you use unless I ask you to check what I found.
Feb 02 05:31:23 <brolin_empey> KotCzarny: You did not mention the price of the converter in your answer.
Feb 02 06:26:12 <KotCzarny> brolin: im on vacations now, does search of 'msata x40' on ebay/ali return something? as for the proce something in 10-20usd range
Feb 02 06:26:27 <KotCzarny> *price
Feb 02 06:51:08 <brolin_empey> KotCzarny: I will check later.
Feb 02 09:29:25 <jonwil> hi
Feb 02 18:10:53 <Vajb> is it easy to add feature to BNF? I would like it to show cycle count since last calibration.
Feb 02 18:48:18 <Sicelo> Vajb: yes, should be possible.
Feb 02 18:49:17 <Vajb> is it possible for user without programming knowledge?
Feb 02 18:51:11 <Sicelo> those are some of the things that made me purge bnf from my system. that said, it is just a script. the value you are looking for is in register 2A of  /sys/class/power_supply/bq27200-0/registers
Feb 02 18:51:30 <Sicelo> sorry
Feb 02 18:51:36 <Sicelo> 28 :)
Feb 02 18:54:33 <Sicelo> so just look at how bnf scripts read values from the registers, and how they display results. should be fairly easy. (dpkg -L bnf will show you where they live)
Feb 02 19:03:57 <Vajb> ok. Will investigate when i get back home, thx
Feb 02 20:49:59 <dayyqx> Why does `ssh` ask me for a password, when I have properly setup auth via id_rsa.pub & authorized_keys?
Feb 02 20:53:10 <kerio> it's probably asking for the passphrase to the key
Feb 02 20:53:40 <kerio> if it's actually asking for the password, check for the permissions to .ssh and to .ssh/authorized_keys
Feb 02 20:53:46 <kerio> first should be 700, second should be 600
Feb 02 20:53:54 <kerio> ssh *is* picky about this
Feb 02 20:54:02 <Sicelo_> hmm, someone had ssh asking password here at some point .. forgot who that was .. it was some simple thing really
Feb 02 20:56:52 <sixwheeledbeast^> maybe an extras package has asked you to set a password? I can think of a few packages that use zenity or the like to set a password
Feb 02 20:58:56 <dayyqx> kerio: i just checked the permissions -- they are correct
Feb 02 20:59:40 <dayyqx> sixwheeledbeast^: it's just straight from the command line
Feb 02 20:59:53 <dayyqx> Sicelo_: can you remember how did it go?
Feb 02 21:00:53 <Sicelo_> i'm looking in logs .. i think it was L29ah
Feb 02 21:01:51 <dayyqx> i'm really just wondering because i have the usual stuff setup: ssh-agent, keys, etc.
Feb 02 21:02:03 <dayyqx> ssh to the device is fine, though.
Feb 02 21:02:33 <dayyqx> Sicelo_: thanks
Feb 02 21:07:14 <Sicelo_> my google-fu fails me
Feb 02 21:07:28 <dayyqx> Sicelo_: that's ok
Feb 02 21:07:34 <Sicelo_> only returns: Sicelo, debug1: Offering RSA public key: /home/sicelo/.ssh/id_rsa, 21:31 ..
Feb 02 21:08:06 <Sicelo_> that's the exact conversation, but no indication as to when. it is within the last month is all i remember
Feb 02 21:08:18 <dayyqx> hmm
Feb 02 21:09:21 <Sicelo_> look like you have problems ssh'ing to other server .. the proble in this case was ssh into n900 .. anyway may still be same cause. if i could find the exact thread
Feb 02 21:10:38 <dayyqx> yes, ssh'ing to n900 is fine, the other way around isn't though. 
Feb 02 21:11:23 <Sicelo_> when adding verbosity with -vvv ... doesn't reveal cause?
Feb 02 21:12:50 <dayyqx> Sicelo: let me grab the output
Feb 02 21:16:48 <Sicelo> maybe also do same on a pc .. and compare differences
Feb 02 21:18:18 <dayyqx> Sicelo: here it is: http://lpaste.net/151505
Feb 02 21:20:34 <NIN101> debug3: Not a RSA1 key file /home/user/.ssh/id_rsa.
Feb 02 21:24:50 <dayyqx> NIN101: I created that file with 'ssh-keygen -t rsa'. I deleted it, then re-created it. It still behaves the same. :-(
Feb 02 21:25:19 <Sicelo> actually i have that line myself, where ssh succeeds
Feb 02 21:26:53 <Sicelo> weird. ... but in the absence of other ideas, purge ssh, remove the ~/.ssh/ directory, and start fresh
Feb 02 21:27:13 <NIN101> dayyqx: are you certain it's not a server side issue?
Feb 02 21:27:43 <dayyqx> NIN101: yup. :)
Feb 02 21:27:48 <dayyqx> Sicelo: let me do that
Feb 02 21:28:03 <Sicelo> asked for output from a pc or other device ;)
Feb 02 21:29:32 <Sicelo> i've got almost exact same output as yours, difference only comes after line 121 ... mine shows server accepting my priv key, yours silently ignores it
Feb 02 21:31:10 <dayyqx> how do you start? I go with: 1- eval $(ssh-agent -s), 2- ssh-add, 3- ssh 
Feb 02 21:32:04 <Sicelo> everytime?
Feb 02 21:32:45 <Sicelo> i've got my public key on the 'server' so i do just `ssh my-pc`
Feb 02 21:33:10 <dayyqx> 1 & 2, initially
Feb 02 21:33:22 <Sicelo> ok. something similar, yes
Feb 02 21:33:37 <dayyqx> the n900's ~/.ssh/id_rsa.pub is on the server's ~/.ssh/authorized_keys
Feb 02 21:36:39 <Sicelo009N> N900 is only device that has ssh?
Feb 02 21:37:08 <dayyqx> Sicelo009N: at the moment, yes. I'll try other machine later.
Feb 02 21:37:27 <dayyqx> *machines
Feb 02 21:38:06 <dayyqx> I remember that I was able to SSH to the server flawlessly with other machines.
Feb 02 21:41:10 <DocScrutinizer05> Vajb: (cyclecount) use ~bq27k-detail2
Feb 02 21:41:42 <Sicelo009N> or, stop the ssh server on your 'server' .. start it manually with sshd -ddd .. that will show you from server side why it fails
Feb 02 21:42:00 <dayyqx> Sicelo009N: let me do that
Feb 02 21:42:21 <Vajb> DocScrutinizer05: im just trying to add that to bnf. So far i managed to break the script lol
Feb 02 21:42:47 <DocScrutinizer05> I dunno bnf
Feb 02 21:43:09 <Sicelo009N> lol, really broken? :P
Feb 02 21:43:12 <Vajb> got it to work
Feb 02 21:43:13 <DocScrutinizer05> I got my bq27k-detail2 and that's just fine for me
Feb 02 21:43:19 <Sicelo009N> good start
Feb 02 21:43:22 <Vajb> i had one space
Feb 02 21:43:29 <Vajb> too much
Feb 02 21:43:43 <DocScrutinizer05> ~bq27k-detail
Feb 02 21:43:43 <infobot> from memory, bq27k-detail is http://maemo.cloud-7.de/maemo5/usr/local/sbin/bq27k-detail2, or ~bq27
Feb 02 21:44:05 <Vajb> CYCL=$ (grep...) had to be CYCL=$(grep...)
Feb 02 21:44:21 <DocScrutinizer05> might eventually upgrade it to also work on /sys/class/power_supply/bq27200-0/registers
Feb 02 21:44:42 <Vajb> that's where bnf pulls the data
Feb 02 21:45:02 <Sicelo009N> are you reading the register? 0x28?
Feb 02 21:45:08 <DocScrutinizer05> yeah, bnf only works with deprecated bq27xx.ko
Feb 02 21:45:24 <Vajb> 0x28 is the value im looking and it is 0x07 atm. I suppose it means sevnth cycle?
Feb 02 21:45:33 <Vajb> or it needs some calculation?
Feb 02 21:45:51 <DocScrutinizer05> cat bq70k-detail
Feb 02 21:45:57 <Vajb> seventh*
Feb 02 21:46:02 <DocScrutinizer05> bq27k-detail even
Feb 02 21:46:37 <DocScrutinizer05> echo "0x29 - 0x28: `pv 0x29 0x28`"   " CYCL Cycle Count Since Learning Cycle High-Low Byte  Cycles       R"
Feb 02 21:46:39 <DocScrutinizer05> echo "0x2B - 0x2A: `pv 0x2B 0x2A`"   " CYCT Cycle Count Total High - Low Byte               Cycles       R"
Feb 02 21:47:18 <Sicelo009N> that means 7, yes
Feb 02 21:47:39 <Vajb> so i have it working the way i want \o/
Feb 02 21:47:40 <Sicelo009N> now format & display it ;)
Feb 02 21:47:47 <DocScrutinizer05> pv doesn't do calculations, so yes, 0x07 means 7
Feb 02 21:48:27 <Vajb> i removed calibration needed value and commented it out
Feb 02 21:48:42 <Vajb> added cycle value there
Feb 02 21:49:00 <Sicelo009N> okay.
Feb 02 21:49:20 <Vajb> CYCL=$(grep 0x28= registers | cut -d= -f2)
Feb 02 21:49:28 <Vajb> that's what pulls it
Feb 02 21:49:28 <DocScrutinizer05> could you please provice a pastebin of `cat /sys/class/power_supply/bq27200-0/registers` ?
Feb 02 21:49:51 <Vajb> ok doc. Gimme a sec
Feb 02 21:50:24 <Sicelo009N> for my needs i found old bq27200 script had all the info i needed, except wall charger detection ... so i 'fixed' it for replacement-bme, and added wallcharger detection .. so i'm happier now :)
Feb 02 21:50:38 <Vajb> Cycle: $CYCL        VDQ:....
Feb 02 21:51:09 <Vajb> that cycle part replaced original "calibration needed"
Feb 02 21:51:35 <Sicelo009N> incidentally, bencoh did exact same stuff i did, but in C instead of script
Feb 02 21:52:05 <DocScrutinizer05> http://maemo.cloud-7.de/maemo5/usr/local/sbin/bq27200.sh
Feb 02 21:52:59 <DocScrutinizer05> if you're interested in perl: http://maemo.cloud-7.de/maemo5/usr/local/sbin/bq27k-detail-perl
Feb 02 21:53:28 <Sicelo009N> DocScrutinizer05: does that use i2c-get?
Feb 02 21:53:39 <DocScrutinizer05> the perl thing? nfc
Feb 02 21:53:53 <Vajb> DocScrutinizer05: pastebin.com/raw/2tAx4DBS
Feb 02 21:53:57 <DocScrutinizer05> ta
Feb 02 21:54:10 <DocScrutinizer05> *cough* ohwell
Feb 02 21:55:48 <DocScrutinizer05> ok, easy, just need to cut off the nonsensical "<addr>=" part
Feb 02 21:57:03 <Vajb> yup
Feb 02 21:57:29 <Vajb> how to cut also 0x part of value?
Feb 02 22:02:06 <DocScrutinizer05> s/0x..=0x//
Feb 02 22:02:15 <DocScrutinizer05> s/^0x..=0x//
Feb 02 22:02:51 <DocScrutinizer05> OH FUCK, it assumes! it assumes 32bit for some registers
Feb 02 22:03:08 <DocScrutinizer05> 0x04=0xffff
Feb 02 22:03:44 <DocScrutinizer05> that's as bad as it gets for a kernel API
Feb 02 22:04:07 <DocScrutinizer05> a kernel API never should assume stuff like that
Feb 02 22:04:16 <dayyqx> Sicelo009N: I found what caused it in the systemd logs: ~/.ssh is a symlink to a dir under ~. Now that dir has the incorrect permissions. :)
Feb 02 22:04:34 <DocScrutinizer05> ohmy
Feb 02 22:04:58 <DocScrutinizer05> CBA to fix my script for /sys/class/power_supply/bq27200-0/registers now
Feb 02 22:05:47 <Sicelo009N> what distro on the pc?
Feb 02 22:07:54 <DocScrutinizer05> the path to do it *if* I were interested, wpould be to sed 's/^0x..=0x//; s/([^ ]{2}/\1 /; s/\n//'
Feb 02 22:08:43 <DocScrutinizer05> wut? ~/.ssh is a symlink? WTF?
Feb 02 22:08:46 <Vajb> i used cut -c8-9 instead of -d= -f2
Feb 02 22:09:03 <Vajb> and got the result i wanted :)
Feb 02 22:10:07 <DocScrutinizer05> Vajb: in http://maemo.cloud-7.de/maemo5/usr/local/sbin/bq27k-detail2 ? that won't fly due to 32bit 'registers'
Feb 02 22:10:24 <DocScrutinizer05> 0x04=0xffff
Feb 02 22:10:55 <DocScrutinizer05> -c8-9  of this is just ff
Feb 02 22:11:11 <DocScrutinizer05> which looks like it works but gives wrong results
Feb 02 22:11:23 <Vajb> ah but im using bnf
Feb 02 22:11:31 <DocScrutinizer05> no diff
Feb 02 22:11:47 <Vajb> can u explain more?
Feb 02 22:12:14 <Vajb> now it prints just 07 tho
Feb 02 22:13:32 <DocScrutinizer05> "-d= -f2" is a parameter set for cut that means "fields are separaetd by '=', use second field".  "-c8-9"  means "use 8th + 9th char". the first will give you 0xffff, the second will give you ff
Feb 02 22:14:30 <DocScrutinizer05> the second one will fail on line "0x04=0xffff" and similar, of your pastebin http://pastebin.com/raw/2tAx4DBS
Feb 02 22:16:15 <Vajb> yes i just read about cut and that's where i got idead to make it use 8th and 9th character
Feb 02 22:16:26 <Vajb> im using it only for line 0x28
Feb 02 22:16:27 <dayyqx> Sicelo009N: NixOS
Feb 02 22:16:52 <Vajb> to show cycle count instead of calibration needed flag
Feb 02 22:17:09 <DocScrutinizer05> sed 's/^0x..=0x//; s/([^ ]{2}/\1 /; s/\n//'  should turn a " 0x00=0x00 \n 0x01=0x40 \n 0x02=0x00 \n 0x04=0xffff \n 0x06=0x4d8 \n " into " 00  40 00 ff ff 4 d8  "
Feb 02 22:17:21 <dayyqx> DocScrutinizer05: yes, it is a symlink? why?
Feb 02 22:17:42 <DocScrutinizer05> ssh will barf up
Feb 02 22:18:02 <dayyqx> DocScrutinizer05: not really.
Feb 02 22:18:29 <dayyqx> dayyqx: as long as  the permissions on the target are correct, it's fine
Feb 02 22:18:48 <DocScrutinizer05> Vajb: (only line 0x28) aah ok
Feb 02 22:19:57 <DocScrutinizer05> ugh
Feb 02 22:20:08 <Vajb> DocScrutinizer05: i know that is supposing that over ten cycles it just changes 0x07 to 0x10
Feb 02 22:20:31 <Vajb> but i'll know till weekend :)
Feb 02 22:20:55 <DocScrutinizer05> sed 's/^0x..=0x//; s/([^ ]{2})[0-9a-f]/\1 /; s/\n//'
Feb 02 22:22:18 <DocScrutinizer05> Vajb: nope, this is an issue about 8bit (0xac) vs 16bit (0x1234) register values in /sys/class/power_supply/bq27200-0/registers
Feb 02 22:22:23 <Maxdamantus> https://gist.github.com/Maxdamantus/ada1412f3ef6ae1c440e
Feb 02 22:22:40 <Maxdamantus> That's my C version, can't remember what the difference was from bencoh's.
Feb 02 22:22:46 <DocScrutinizer05> it doesn't matter if cycle count changes from 0x07 to 0xff
Feb 02 22:24:46 <Maxdamantus> It uses strtol (though converts the results to ints), so probably doesn't have this issue.
Feb 02 22:25:54 <bencoh> hm ?
Feb 02 22:26:06 <bencoh> my version of what ?
Feb 02 22:26:11 <DocScrutinizer05> in bq27k-detail2 replace >> dmppath=`find /sys/devices/ -path '*/hdq/*' -name dump` << by >> dmppath=`find /sys/devices/ -path '*/bq27200*/*' -name registers` <<
Feb 02 22:26:57 <Maxdamantus> doc mentioned your C version of bq27200.sh .. hmm .. maybe he was just thinking of mine, thought I vaguely recalled someone else doing something similar though.
Feb 02 22:27:16 <DocScrutinizer05> in bq27k-detail2 replace >> dmpcmd="cat ${1:-${dmppath}}| tr -s '\n ' '  '"  << by >> dmpcmd="cat ${1:-${dmppath}}| sed 's/^0x..=0x//; s/([^ ]{2})[0-9a-f]/\1 /; s/\n//'  <<
Feb 02 22:27:58 <bencoh> Sicelo009N: did I write some bq27-related code?
Feb 02 22:28:43 * bencoh $ fsck /dev/brain/mem
Feb 02 22:35:27 <dayyqx> Sicelo009N: thanks for the assistance. I'm parting.
Feb 02 22:38:28 <DocScrutinizer05> oh shit, it even omits leading zeroes
Feb 02 22:38:46 <DocScrutinizer05> how braindamaged is THAT?
Feb 02 22:39:36 <DocScrutinizer05> 0x06=0x4d8
Feb 02 22:39:38 <DocScrutinizer05> 0x08=0xe3c
Feb 02 22:39:39 <DocScrutinizer05> 0x0a=0x04
Feb 02 22:40:13 <DocScrutinizer05> this "registers" sysnode is as borked as it may get
Feb 02 22:40:50 <DocScrutinizer05> if that's the mandated format for any "registers" sysnodes in linux, then I prolly switch to windows
Feb 02 22:43:03 <DocScrutinizer05> silly enough it does NOT omit leading zeroes in 2digit values like 0x28=0x07
Feb 02 22:44:44 <DocScrutinizer05> this is definitely broken, a bug
Feb 02 22:47:55 <DocScrutinizer05> even if we ignore (and tolerate) assumptions about registers being 8bit or 16bit, it *always* should be FOUR digits for 16bit registers, like it's always TWO digits for 8bit registers and values <=0xFF (for the latter it should be 4 difgits no matter which value, for those 16 bit regs. Or rather it shouldnt make assumptions of 2 8bit registers forming a 16bit value, to start with)
Feb 02 22:48:19 <DocScrutinizer05> there are no 16bit registers on I2C
Feb 02 22:49:01 <DocScrutinizer05> Pali: ^^^ bug
Feb 02 22:49:12 <DocScrutinizer05> Pali: in bq27xx.ko
Feb 02 22:51:26 <Pali> DocScrutinizer05: this is kernel-power on maemo right?
Feb 02 22:51:36 <DocScrutinizer05> prolly, yes
Feb 02 22:51:42 <DocScrutinizer05> ask Vajb
Feb 02 22:51:58 <DocScrutinizer05> http://pastebin.com/raw/2tAx4DBS
Feb 02 22:52:00 <Pali> just leading zeros are missing
Feb 02 22:52:43 <DocScrutinizer05> yes, it should be %04x at very least. Though actually I2C registers are *always* just 8bit
Feb 02 22:53:21 <Pali> those are bq registers and some are really 16bit
Feb 02 22:53:38 <DocScrutinizer05> no, there are technically no 16bit registers on I2C
Feb 02 22:54:18 <DocScrutinizer05> assuming two registers on I2C form a 15bit value is just that: an assumption, which is not appropriate for kernel API level
Feb 02 22:54:29 <DocScrutinizer05> 16bit even
Feb 02 22:55:42 <Maxdamantus> I think it's probably weirder that thece are a bunch of null bytes at the end of the file.
Feb 02 22:55:53 <DocScrutinizer05> such high level 'assumtions' are for userland accessing and interpreting the register dump
Feb 02 22:56:58 <bencoh> 00:53 < DocScrutinizer05> no, there are technically no 16bit registers on I2C
Feb 02 22:57:11 <bencoh> hm? are you talking in this specific case ?
Feb 02 22:57:26 <DocScrutinizer05> no, this is a question of I2C protocol
Feb 02 22:57:47 <DocScrutinizer05> I2C defines a max 128 registers of 8 bit each
Feb 02 22:57:59 <bencoh> wtf?
Feb 02 22:58:40 <bencoh> well, i2c (in its most basic form) defines 7b addresses, and 8bit words
Feb 02 22:58:56 <bencoh> but your frame can be as long as you want
Feb 02 22:59:59 <bencoh> as long as you keep ACKing after every read (for read accesses)
Feb 02 23:00:13 <Vajb> 2.6.28.10-power53 if my kernel still matters
Feb 02 23:00:56 <DocScrutinizer05> bencoh: what I just said. 128 registers a 8bit per register
Feb 02 23:00:59 <Vajb> and cssu-testing
Feb 02 23:01:05 <DocScrutinizer05> since addr is 7bit
Feb 02 23:01:09 <DocScrutinizer05> http://wstaw.org/m/2016/02/03/plasma-desktopPv3616.png
Feb 02 23:02:02 <DocScrutinizer05> interpretation of register semantics is in named sysnodes, not */registers
Feb 02 23:02:32 <DocScrutinizer05> */registers should list up to 128 plain 8bit values
Feb 02 23:02:48 <DocScrutinizer05> as 0xff
Feb 02 23:02:57 <DocScrutinizer05> aka %02x
Feb 02 23:03:45 <DocScrutinizer05> */rehisters is exactly for those usecases where you do _not_ want any semantics applied
Feb 02 23:03:59 <bencoh> DocScrutinizer05: this schema only refers to the case where your device defines 8b-addressed 8b registers (?)
Feb 02 23:04:18 <DocScrutinizer05> that's I2C spec
Feb 02 23:04:21 <bencoh> no
Feb 02 23:04:23 <DocScrutinizer05> bencoh: ^^^
Feb 02 23:04:31 <bencoh> at least not that I know of
Feb 02 23:04:35 <DocScrutinizer05> and it's 7bit addr
Feb 02 23:04:45 <DocScrutinizer05> 8th bit is r/w
Feb 02 23:04:49 <bencoh> the slave address is 7b
Feb 02 23:04:54 <bencoh> not the reg address
Feb 02 23:04:57 <DocScrutinizer05> oops, ok
Feb 02 23:05:00 <bencoh> (in your example)
Feb 02 23:05:11 <DocScrutinizer05> then it's 256 registers max
Feb 02 23:05:25 <bencoh> i2c eeproms I know of have 16b addresses
Feb 02 23:05:28 <bencoh> for 8b words
Feb 02 23:05:55 <ds3> hmmm?
Feb 02 23:06:00 <DocScrutinizer05> well, I wonder how the I2C interface controllers will handle those
Feb 02 23:06:13 <bencoh> the i2c spec doesn't say anything about the data you transfer
Feb 02 23:06:29 <bencoh> it only specifies the "how" you transfer it
Feb 02 23:06:30 <ds3> there are 16bit register devices
Feb 02 23:06:36 <bencoh> ds3: sure there are :-)
Feb 02 23:06:43 <ds3> and they don't work if you read them as 8 bit values
Feb 02 23:06:52 <DocScrutinizer05> aha
Feb 02 23:07:08 <DocScrutinizer05> anyway bq27k isn't one of those
Feb 02 23:07:26 <bencoh> DocScrutinizer05: anyway, time to re-read the bq27x ref I guess, to see if this really is a bug or not
Feb 02 23:07:33 <ds3> the kernel API doesn't really care 
Feb 02 23:08:13 <DocScrutinizer05> the kernel api for I2C is like "read N bytes, starting at X"
Feb 02 23:08:49 <bencoh> DocScrutinizer05: well, it's quite close to what you'd need for most devices
Feb 02 23:08:58 <bencoh> as long as you can specify the address length as well
Feb 02 23:08:58 <DocScrutinizer05> so what?
Feb 02 23:09:31 <bencoh> (read N bytes starting at address X of length n)
Feb 02 23:09:38 <ds3> that must be the simplified one
Feb 02 23:09:43 <bencoh> probably yeah
Feb 02 23:09:58 <DocScrutinizer05> it's not the kernel driver's job to apply 'knowledge' about 2 8bit register addr forming a 16bit value, ** IN */registers **
Feb 02 23:09:59 <ds3> the full API is - WRITE n BYTES READ x BYTES STOP
Feb 02 23:10:41 <bencoh> ds3: yeah but most people dont want to handle the waiting :)
Feb 02 23:10:44 <DocScrutinizer05> and I'm not discussing the I2C function call interface here
Feb 02 23:10:58 <ds3> bencoh: waiting? you mean you the async nature of that interface?
Feb 02 23:11:23 <DocScrutinizer05> it's about /sys/class/power_supply/bq27200-0/registers
Feb 02 23:12:13 <bencoh> ds3: be it spinlock or async/interrupt-driven :)
Feb 02 23:12:34 <bencoh> DocScrutinizer05: so you're saying it's stripping leading zeros?
Feb 02 23:12:37 <DocScrutinizer05> this chip evidently addresses 8bit registers, and some of them form pairs for a 16bit value, however that doesn't beling into /sys/class/power_supply/bq27200-0/registers
Feb 02 23:12:48 <DocScrutinizer05> bencoh: yes, that too
Feb 02 23:13:17 <DocScrutinizer05> http://pastebin.com/raw/2tAx4DBS
Feb 02 23:13:30 <DocScrutinizer05> 0x08=0xe3c
Feb 02 23:13:31 <bencoh> This page has been removed!
Feb 02 23:13:45 <DocScrutinizer05> mmmphh
Feb 02 23:14:21 <DocScrutinizer05> http://paste.opensuse.org/81405177 hgere you are
Feb 02 23:15:33 <DocScrutinizer05> 0x16=0x7d
Feb 02 23:15:34 <DocScrutinizer05> 0x18=0xffff
Feb 02 23:17:01 <DocScrutinizer05> 0x16 is a "16 bit register" (though I really think it should be 2 8bit registers), you can tell from next addr being 0x18. Yet it has only 2 digits (0x7d), with a different value it would have 3 or 4 digits, but silly enough never 1 digit only
Feb 02 23:17:45 <bencoh> ?
Feb 02 23:18:07 <DocScrutinizer05> 0x28=0x07
Feb 02 23:18:09 <DocScrutinizer05> 0x2a=0x23b
Feb 02 23:18:25 <DocScrutinizer05> why isn't it 0x28=0x7  ?
Feb 02 23:18:39 <bencoh> leading zeros when you're reading from /sys nodes don't really matter; unless you want to feed it back
Feb 02 23:18:45 <DocScrutinizer05> anyway it SHOULD be 0x28=0x0007
Feb 02 23:18:48 <bencoh> (?)
Feb 02 23:20:44 <DocScrutinizer05> well, I suggest the format "ADRESS twentyeight HEX IS VALUE 7 HEX" then ;-)  When format doesn't matter
Feb 02 23:20:59 <DocScrutinizer05> literally
Feb 02 23:21:16 <DocScrutinizer05> the idea of sysnodes is to allow easy machine parsinfg
Feb 02 23:21:43 <DocScrutinizer05> emphasis on "easy"
Feb 02 23:23:03 <DocScrutinizer05> and particularly */registers should allow me to read out value of register 0x29 (SIC!) no matter which semantics the register might have
Feb 02 23:25:54 <bencoh> it doesn't let you read 0x29?
Feb 02 23:25:56 <DocScrutinizer05> so the above should read 
Feb 02 23:25:57 <DocScrutinizer05> 0x28=0x07
Feb 02 23:25:58 <DocScrutinizer05> 0x29=0x00
Feb 02 23:26:00 <DocScrutinizer05> 0x2a=0x02
Feb 02 23:26:01 <DocScrutinizer05> 0x2b=0x3b
Feb 02 23:26:18 <DocScrutinizer05> do you see any 0x29 in http://paste.opensuse.org/81405177 ?
Feb 02 23:26:40 <bencoh> I don't, that's why I'm asking :)
Feb 02 23:27:11 <bencoh> but, okay, I get your point :)
Feb 02 23:27:28 <DocScrutinizer05> exactly the point. There's evidently a register 0x29 in bq27200, however it's not really available in /sys/class/power_supply/bq27200-0/registers
Feb 02 23:27:36 <bencoh> it's more a representation issue than an internal bug
Feb 02 23:27:46 <bencoh> (I think)
Feb 02 23:28:39 <DocScrutinizer05> yeah, the bug is in printf() and some logic that concatenates registers into one value despite */registers is meant to show a raw dump without any semantics applied
Feb 02 23:29:50 <DocScrutinizer05> when you start applying semantics there in */registers, then why the heck are the values hex and not signed or unsigned integers already?
Feb 02 23:31:51 <DocScrutinizer05> the named sysnodes like */ttf are supposed to apply all the needed semantics to provide something "human readable". */registers in my book is a *raw* dump of *all* existing registers of the chip
Feb 02 23:33:48 <DocScrutinizer05> since generaly when accessing */registers you want to bypass the semantics in kernel driver
Feb 02 23:36:22 <DocScrutinizer05> there are even chip that have register banks, where semantic of register M depends on the bank enmabled by writing a value to e.g. reg 0x01
Feb 02 23:37:11 <DocScrutinizer05> I don't think you should expect changing format in */registers sysnode for such chips, depending on current value of reg 0x01
Feb 02 23:38:14 <DocScrutinizer05> as a rule of thumb, a tool should be able to handle */registers of arbitrary chips in a uniform way
Feb 02 23:39:53 <DocScrutinizer05> btw same applies for all sysnodes. That's the reason why */current is in micro-Amperes
Feb 02 23:40:07 <DocScrutinizer05> or was it even nano-Amperes?
Feb 02 23:41:41 <DocScrutinizer05> I really hope */registers isn't specified in linux for all kernel drivers to look like bq27xx.ko's looks now, 
Feb 02 23:56:07 <sunshavi> Hi guys, any of you is using maemo-skd with scratchbox 2?
**** ENDING LOGGING AT Wed Feb 03 02:59:58 2016