2015-03-02 GnuCash IRC logs

00:15:11 <bozonius> night
00:15:16 *** bozonius has quit IRC
00:40:31 *** O01eg has quit IRC
01:10:23 *** Krzysiek_K has joined #gnucash
01:19:21 *** MechtiIde has joined #gnucash
01:26:27 *** cartsoftware1 has joined #gnucash
01:26:28 *** cartsoftware has quit IRC
02:00:51 *** keret has joined #gnucash
02:09:29 <keret> recording stock transactions with securities that have had either a name change or a code change. any guidance on how best to handle these?
02:09:39 *** MechtiIde has quit IRC
02:10:58 <keret> would it be best to record these adjustments using basic journal entries to record the disposal from old account to new account? or simply change the existing transactions to include the new account?
02:19:17 *** cartsoftware1 has quit IRC
02:28:57 *** Krzysiek_K has quit IRC
02:52:46 <kimmo2_> I would probably just transfer the assets from one account to another
03:03:35 *** gjanssens has joined #gnucash
03:03:35 *** gncbot sets mode: +o gjanssens
03:03:45 *** pacon has joined #gnucash
03:15:58 *** cartsoftware has joined #gnucash
03:45:47 <keret> kimmo2_: the more i think about it, that is probably the simplest method. can also add a relevant note/memo.
03:45:56 <keret> although reporting based on historical prices will be affected. is there any simple method to transfer prices from one security to another?
03:46:06 *** cartsoftware has quit IRC
03:58:41 <kimmo2_> hmm, that would probably require transferring them in separate blocks, one block for each purchase
03:59:15 <kimmo2_> then again, I do my stocks accounting based on purely the purchase price, with my books knowing nothing about the quantity or price of individual purchases
04:01:14 <keret> would definitely have to retain the various lots. can also be complicated by partial sales of lots over time. i guess these are complications only if using gnc to track stock quantities - ie using accounts with type "stock".
04:02:18 <keret> thinking through a few of the issues today, it seems it can be a little more complicated than initially thought.
04:02:50 <keret> perhaps ideally securities would retain a history of name/code changes and this history would propogate through any historical reporting.
04:14:14 *** cartsoftware has joined #gnucash
04:16:39 *** fabior has joined #gnucash
04:35:35 *** cartsoftware has quit IRC
04:46:35 *** StuM has quit IRC
04:52:20 *** cartsoftware has joined #gnucash
05:01:43 *** wol has joined #gnucash
05:11:39 *** uXus has quit IRC
05:11:49 *** wol has quit IRC
05:11:54 *** wol has joined #gnucash
05:14:57 *** uXus has joined #gnucash
05:15:28 *** mohammad has joined #gnucash
05:15:51 <mohammad> i like gnucash but i need to it in Persian language what should i do?
05:16:33 <mohammad> could someone help me here please?
05:17:35 <Simon> keret: if it's still the same stock, surely you could just change the name and code?
05:18:29 <Simon> mohammad: http://wiki.gnucash.org/wiki/Translation
05:19:29 <Simon> according to the list of translations, Persian is already supported
05:19:56 <Simon> at least partially... the translationproject website is unclear
05:21:52 <mohammad> Simon: should i do translation myself?
05:22:13 <Simon> well, if there are missing phrases and you're able to translate it...
05:22:50 <Simon> http://translationproject.org/domain/gnucash.html it looks like only 25% of the messages are translated
05:23:01 <mohammad> Simon: where can i see translated languages?
05:26:39 <mohammad> Simon: it is 1612 of 4569
05:33:40 *** cartsoftware has quit IRC
05:33:50 *** cartsoftware has joined #gnucash
05:43:54 *** cartsoftware has quit IRC
05:44:57 *** cartsoftware has joined #gnucash
05:55:00 <keret> Simon: agreed. that's one option. change the name/code in the security editor, change the name/code of the stock account and change the name of any trading account, if using trading accounts. if the entire history of transactions in the relevant stock is updated, any historical reports that reference the stock by the old name will no longer be supported by the underlying data.
05:57:14 <keret> not sure what the strictly correct accounting procedure is.
06:02:42 <keret> or how gnc deals with this issue as compared to what would be considered best practice.
06:05:49 <keret> also unsure as to whether there are any other issues to consider particular to how gnc works (as i've only just started using it). haven't been using it for very long).
06:35:59 *** GabrieleV_ has joined #gnucash
06:36:04 *** GabrieleV has quit IRC
06:36:04 *** GabrieleV_ is now known as GabrieleV
06:41:13 *** himaxx has joined #gnucash
06:57:20 *** Jimraehl1 has left #gnucash
06:57:46 *** Jimraehl1 has joined #gnucash
07:09:58 *** pacon has quit IRC
07:10:14 *** fabior has quit IRC
07:16:13 *** cartsoftware has quit IRC
07:38:57 *** Hinnerk has joined #gnucash
07:39:14 <Hinnerk> Hi
07:40:42 <Hinnerk> I still don't understand what's going wrong with Bayes sometimes.
07:41:16 <Hinnerk> I have a monthly transaction going through my checking account, which always looks the same and its other leg is always supposed to go into the same account.
07:41:23 <Hinnerk> Still, it is never recognised.
07:41:48 <Hinnerk> I verified that there are no "old Versions" of this trx, possibly being booked elsewhere.
07:42:55 <Hinnerk> same with a few other ones.
07:43:18 <Hinnerk> I works fine for most - but sometimes it just doesn't get the proper choice when it should be simple.
07:43:39 <Hinnerk> Is the algo looking at the entire string or just the first n chars?
07:44:24 <Hinnerk> Because some descriptions are quite long, starting with reference and mandates etc. which is not very unique from a pattern perspective.
07:44:48 <Hinnerk> But the clear hints, e.g. company name, always follow later in the string.
07:45:30 <warlord> keret: or just change the security name
07:46:09 <Hinnerk> Additional question: Is the algo doing all the work on the fly, or is it building a "database" in the file? If so, maybe a "reset" would be usefull.
07:46:54 <warlord> Hinnerk: it builds a database, stored in the KVP of the account.
07:47:03 <Hinnerk> KVP?
07:49:48 <keret> warlord: name changes are relatively simple. update the name in the security editor and account register, wherever necessary.
07:53:31 <keret> code changes seem a little more complicated. update the code in the security editor and account register, including the name of the trading account used. but like i mentioned earlier, would all portfolio reports run for prior periods now show the new name/code even though at that time the security was named something else or had an older code?
07:56:23 *** rickoehn has joined #gnucash
07:56:57 <keret> i haven't tested, but what about prices? are they linked to the underlying security in such a way that any name/code change will not affect a given security being able to use a related price series?
08:00:04 <warlord> yes, it would give the new name
08:00:26 <warlord> prices.... I dont recall if the price is based on the commodity GUID or symbol.
08:03:09 <keret> thx. will play around further and see how the prices are dealt with.
08:05:21 <keret> part of the issue is that i imported a bunch of stock transactions, some involving the same stock with old/new names/codes.
08:05:25 <keret> in hindsight, easier if i import the transactions for old stock name/code first, adjust name/code, then import transactions for new stock name/code.
08:05:53 <keret> or adjust names/codes of old stock to match new name/code before importing.
08:07:31 <warlord> yeah.
08:07:40 <warlord> There are different ways to handle it.
08:08:07 *** lmat has joined #gnucash
08:08:09 <warlord> how you handle it will depend a lot on how you want to handle the transition. Do you want to treat it as one stock, or two?
08:08:55 <keret> ideally treat it as one stock for tax (lot) purposes but 2 for reporting purposes. that may not be possible.
08:09:08 <warlord> That's not possible
08:09:50 <keret> put another way, treat it as it is treated in reality. again, not possible.
08:11:37 *** fabior has joined #gnucash
08:12:52 <keret> on another matter, any thoughts/guidance on handling scrip-for-scrip takeovers (ie shareholder receives stock in acquiring company based on shareholding in target company). unsure if same term used in US.
08:13:59 <keret> perhaps similar to code change but may have some adjustment in holding (eg 1 share in newco issued for every 2 shares in oldco).
08:15:28 <keret> perhaps a combination of a code/name change and a stock split/merger?
08:20:42 *** himaxx has joined #gnucash
08:23:13 *** wol has quit IRC
08:24:59 *** himaxx has quit IRC
08:27:26 <Hinnerk> any idea how/if I can fix my Bayes-database?
08:29:19 *** lmat has quit IRC
08:30:25 <warlord> keret: I would treat that as two different companies and just a transfer from OldCo to NewCo
08:30:30 <warlord> Hinnerk: there is no UI method to reset it. Backup your data file and remove it by hand.
08:30:38 <Hinnerk> how will it build up over time? And how do i recognise the proper sections?
08:31:33 <keret> warlord: "transfer from OldCo to NewCo". how would you handle this?
08:32:20 <warlord> keret: credit Assets:Stocks:OldCo, debit Assets:Stocks:NewCo -- the same way you would transfer from Checking to Savings.
08:34:10 <Hinnerk> is it this section of "slot:key" and "slot:value"?
08:34:53 *** lmat has joined #gnucash
08:41:07 <keret> warlord: ok. where i am the acquisition date and cost base can be rolled over from oldco stock to newco stock for tax purposes so presumably i would need to do this for each remaining oldco lot using the original acqusition date. cost base remains the same with price adjusted for the takeover terms offered.
08:42:10 <warlord> Hinnerk: yes. make sure you only delete the bayes data
08:42:45 <warlord> keret: yes; for 'purchase date' you kind of need to keep track of that.. separately.
08:46:52 <keret> probably just create a "transfer" entry for each lot with using the original purchase date with a memo/note confirming reason for the transfer, although this will probably play havoc with any historical reporting. tricky!
08:48:49 *** aqua___ has joined #gnucash
08:50:57 <keret> ok, think i misunderstood what you meant by your last comment, but can see now that setting the purchase date as the date of transfer may be necessary.
08:52:24 <Hinnerk> I was thinking of just hand-pruning it.
08:52:29 <Hinnerk> what do I need to know?
08:53:34 <keret> warlord: earlier you said: "how you handle it will depend a lot on how you want to handle the transition. Do you want to treat it as one stock, or two?" can you elaborate on how you treat the 2 alternatives? if treat as 1 stock, change security name/code? if treat as 2 stocks, transfer from account A to account B?
08:55:12 <Hinnerk> I also lost track of where these slots are again - the file is huge, aber 500'000 lines...
08:55:31 <Hinnerk> what can I enter as search in my text editor?
08:55:36 <Hinnerk> slot seems to be too general.
08:57:57 <warlord> keret: yes
08:57:58 *** aqua___ has quit IRC
08:58:28 <warlord> Hinnerk: find the account first.
09:03:32 *** gncbot has joined #gnucash
09:08:20 *** warlord sets mode: +o gncbot
09:08:57 <warlord> mohammad: you need to install all the build dependencies; what OS/Distro are you using?
09:09:26 <mohammad> warlord: linux mint 17 qiana
09:09:44 <warlord> is that an apt-based system?
09:09:51 <mohammad> warlord:yes
09:09:56 <warlord> try: apt-get build-dep gnucash
09:10:44 <mohammad> warlord: E: Unable to find a source package for gnucash
09:11:34 <warlord> mohammad: Umm.... I suspect that means you don't have the correct package/sources list(s) defined. Alas, you've reached hte limit of my knowledge (I run Fedora here)
09:11:46 <warlord> But basically, you'll need to install all the package dependencies.
09:11:51 <warlord> Try looking at README.dependencies
09:12:21 <mohammad> warlord: i can install gnucash by running sudo apt-get install gnucash
09:12:54 <mohammad> warlord: but i need to compile source code for translation language
09:15:11 <warlord> you can install gnucash but cant find its source package? That really means you dont have the correct source repository listed inyour packages list.
09:15:55 <mohammad> i dont know
09:16:26 <mohammad> warlord: but i installed that and software was great
09:17:39 <warlord> like I said, you'll need to install all the build deps. easiest way is via the package manager. but alternately, look in README.dependencies and manually install all the deps. The error you posted implies you don't even have a basic build environment on your system..
09:18:02 <warlord> You dont have aclocal, automake, or autoconf.. Do you even have the compiler installed?
09:18:45 <warlord> biab..
09:19:39 *** lmat has quit IRC
09:22:21 *** ErKa has joined #gnucash
09:23:09 <mohammad> warlord: i have not had
09:23:15 *** fabior has quit IRC
09:23:36 <mohammad> warlord: i tried just gcc and javac
09:24:40 *** cartsoftware has joined #gnucash
09:25:20 *** lmat has joined #gnucash
09:25:40 *** fabior has joined #gnucash
09:33:02 *** himaxx has joined #gnucash
09:35:52 *** himaxx has quit IRC
09:40:26 *** ErKa has quit IRC
09:54:44 *** jimvideo has joined #gnucash
09:57:56 <warlord> mohammad: okay, so my suggestion remains: either go through README.dependencies and install everything by hand, or figure out what you need to fix in your packages.list to enable the gnucash source package so apt-get build-dep works.
09:58:43 <Hinnerk> woarlord: which account?
09:58:48 <Hinnerk> I found accounts, plenty of them.
09:59:25 <Hinnerk> but how can I lcearly identify the items which refer to Bayes'?
10:02:39 <Hinnerk> For example, I find the account for payable amounts for my electrcity company.
10:02:48 <Hinnerk> But I don't see any entry for Bayes...
10:03:15 <Hinnerk> which of course would explain why GC isn't recognising new trx
10:03:24 <Hinnerk> but why were they never added?
10:04:07 <Hinnerk> And say I know for sure what the trx looks like, how would I add an entry to properly force it to do as I want?
10:04:56 <Hinnerk> hups, typo in names is always very impolite - sorry, warlord.
10:06:43 *** mohammad has quit IRC
10:17:13 <warlord> Hinnerk: it would be the import account..
10:17:20 <warlord> Do you have Bayes turned on?
10:17:30 <warlord> (I dont recall the KVP key used for the bayes database)
10:18:45 *** ErKa has joined #gnucash
10:25:53 *** aqua___ has joined #gnucash
10:28:24 *** wol has joined #gnucash
10:46:31 *** himaxx has joined #gnucash
10:53:30 *** O01eg has joined #gnucash
10:55:14 *** jralls_afk is now known as jralls
11:08:55 *** aqua_ has joined #gnucash
11:08:55 *** aqua___ has quit IRC
11:24:37 *** aqua_ has quit IRC
11:29:43 <jralls> warlord, gjanssens, fell: I'm travelling again, this time to Japan. I'll be there until the 28th. Email at least daily and I may pop in here from time to time.
11:30:59 <fell> Have fun, jralls!
11:32:30 <jralls> Thanks!
11:32:41 *** jralls has quit IRC
11:48:33 *** aqua___ has joined #gnucash
12:01:07 <warlord> safe travels
12:11:35 <perth> OK, So I've got my qif ready, all 20.5K lines of it :), and it fails to import. I'm fairly confident the issue is some special character in a description/memo somewhere. Any thoughts on characters to avoid? file is also in utf-8 not ASCII, but (excluding some special characters?) doesn't seem to matter...
12:12:07 *** MechtiIde has joined #gnucash
12:16:04 *** aqua___ has quit IRC
12:18:01 <Hinnerk> yes, I do have Bayes turned on with all the default thresholds.
12:21:31 <warlord> Hinnerk: then you need to find the import account, and then within that import account would be the bayes database.
12:21:59 <Hinnerk> ok, so I found for my checking account the following slot:
12:22:07 <Hinnerk> <slot:key>import-map-bayes</slot:key>
12:22:12 <Hinnerk> that's it, right?
12:22:23 <Hinnerk> now, Im wondering about some entries:
12:22:44 <Hinnerk> <slot:key>&amp;</slot:key>
12:22:59 <warlord> Hinnerk: that looks right.
12:23:08 <warlord> that would be tokenizing on the string "&"
12:23:11 <Hinnerk> does this really mean it maps trx with an ampersand to the following accounts?
12:23:21 <warlord> e.g., if you have "Marks & Spencers" ....
12:23:29 <Hinnerk> and why does it give 4 different accounts?
12:23:42 <warlord> because that token was used to map to 4 different accounts.
12:23:48 <warlord> Each token gets its own set of values
12:23:56 <perth> :) Forgot the !Type:Bank :)
12:24:13 <warlord> so for "Marks & Spencers" you would have 3 tokens: "Marks", "&", and "Spencers"
12:24:33 <warlord> Each token gets weighted on various accounts, and the final will be the "best match" from combining the tokens.
12:24:52 <Hinnerk> what does the integer mean?
12:25:04 <Hinnerk> <slot:value type="integer">2</slot:value>
12:25:06 <warlord> For example you might also have "Spencers Gifts", which would also map "Spencers" to something.
12:25:17 <warlord> What's the slot:name?
12:25:23 <warlord> Or key..
12:25:39 <Hinnerk> the corresponding key is one of the accounts being mapped to
12:26:10 <warlord> Oh, right.. That's how it's done. (It's been a LONG LONG time since I looked at that code)
12:26:23 <warlord> It's the bayesian value.
12:26:39 <warlord> If you just want to reset it then just remove the whole tree
12:26:48 <warlord> Don't attempt to prune it manually
12:26:54 <Hinnerk> why not?
12:27:07 <Hinnerk> If I know a certain token is unique, why not use?
12:27:29 <Hinnerk> higher value means more meaningfull?
12:27:49 <warlord> I dont recall.
12:27:54 <warlord> You'd have to look in the code.
12:28:20 <Hinnerk> if it corresponds to the threshold, then higher is better.
12:30:21 <Hinnerk> if I delete all, I need to do all the mapping initially manually, right? I won't draw from existing trx?
12:31:55 <Hinnerk> some tokens just seem crazy.
12:32:08 <Hinnerk> there are so many, which contain a certain date!
12:32:35 <Hinnerk> those must have been used in one specific month only...
12:35:31 *** cartsoftware has quit IRC
12:36:01 <Hinnerk> let me give you an example, where it should be quite clear on how to enter a mapping: Some trx containt the name of the "mandate" which allows the creditor to withdraw money from my checkings account.
12:36:10 <Hinnerk> This number is unique.
12:36:14 <Hinnerk> It definitely is.
12:36:34 <Hinnerk> It can always be mapped to my corresponding creditor account.
12:36:38 <warlord> yes, if you delete it all you need to map from the start.
12:37:04 <warlord> I honestly dont recall what the number means. I *think* higher is better, but I wont swear to it.
12:37:45 <Hinnerk> that would be horrible - and it wouldn' guarantee that is work, since I can't find a token corresponding to my electricty company, even though I mapped to it serveral times.
12:39:08 <Hinnerk> Is it really unsafe to add proper lines?
12:39:31 <Hinnerk> SAy if I stick to really unique tokens?
12:46:42 <Hinnerk> Ok, I can empirically confirm that higher numbers are better: I found one unique token, that I one mapped accidently to a wrong account early on. It has two slot: The good one (14) and the bad one (1).
12:47:19 <Hinnerk> oh, even more interesting:
12:47:45 <Hinnerk> The underlying accounts don't exist any more the the token actually is linked to the one that never gets mapped.
12:48:13 <Hinnerk> It seems that changes in the accounts are never noticed by Bayes.
12:48:54 <Hinnerk> is it then at least safe to delete tokens that are outdated?
12:49:07 <Hinnerk> refering to accounts which don't exist anymore?
12:49:47 <Hinnerk> is is correct that the tokens have to be ordered alphabetically?
12:50:18 *** wol has quit IRC
13:02:20 *** Krzysiek_K has joined #gnucash
13:05:44 <warlord> I think so.
13:05:56 *** lmat has quit IRC
13:06:18 <Hinnerk> Ok, I'll give it a shot, unless you give me a hard reason not to manually prune it :)
13:07:14 <warlord> So long as you back a backup file ;)
13:07:31 <Hinnerk> thank you for your vote of confidence :)
13:07:43 <Hinnerk> now I see why you don't want me to...
13:08:09 <warlord> it's very easy to shoot yourself in the foot.
13:08:12 *** lmat has joined #gnucash
13:08:15 <warlord> it's nothing personal
13:08:23 <Hinnerk> I'm just kidding.
13:12:44 *** aqua___ has joined #gnucash
13:33:52 <perth> Sometimes I hate my bank, process some ATM transactions in 2 days, others in 10+
13:53:10 <Hinnerk> man, I can delete so many entries that are either:
13:53:18 <Hinnerk> -refering to accounts that don't exist anymore
13:53:35 <Hinnerk> -use tokens which make sense as a one-time-shot only
13:53:52 <Hinnerk> -use tokens which have really low meaning "00"
13:54:13 <Hinnerk> and usually the meaningfull tokens follow later anyway!
13:56:05 <warlord> Hinnerk: sounds like we could use a "cleanup" function in there.
13:56:20 <Hinnerk> yes, seriously!
13:56:31 <Hinnerk> at least for old accounts.
13:56:37 <Hinnerk> maybe I'm a bit extremely.
13:56:51 <Hinnerk> since I started using gnucash I always used the same file
13:56:59 <Hinnerk> and I have changed my CoA quite a bit
14:01:43 <Hinnerk> has anyone tested not allowing tokens with less than 4 chars?
14:02:03 <Hinnerk> I guess that may not work for everyone, but from that I see, those are usually totally useless.
14:02:57 <Hinnerk> a combination of "009" may have been unique so far - but there are much more trustworthy strings contained. Nothing should be booked based on the "009" appearing somewhere...
14:03:14 <Hinnerk> I guess it is using that just as one piece of a puzzle, but still...
14:06:43 <Hinnerk> and maybe it should split tokens more intelligently.
14:07:09 <Hinnerk> for e.g. I have income coming in with the following token: "id/monthyear"
14:07:20 <Hinnerk> e.g. 123456/052014
14:07:25 <Hinnerk> the id never changes.
14:07:38 <Hinnerk> month and year always do
14:07:51 <Hinnerk> if it used "/" as a separator for tokens, this would help
14:08:13 <Hinnerk> but apparently/empirically it uses only spaces.
14:16:02 <warlord> yes, it only uses spaces, because a slash might still be relevant and *not* be a token separator
14:16:38 <warlord> As for the cleanup code, patches always welcome ;)
14:22:27 <Hinnerk> well, I'm just starting to look at reports - patches will still be in the far, far future...
14:23:20 <Hinnerk> warlord: Would it not be possible to use "/" as a token and a separator?
14:24:39 <Hinnerk> and a token of "01" is just so meaningless...
14:28:23 <warlord> Hinnerk: but that's the wonder of it; by itself it is meaningless, but when it is combined with other tokens it's not.
14:28:59 <Hinnerk> well, looking at the list of tokens, I have my doubt.
14:29:17 <Hinnerk> the token "01" has really low values everywhere and a long list of accounts.
14:29:24 <Hinnerk> each account has more meaningfull tokens.
14:33:35 <warlord> Sure, but you're a human, not a computer ;)
14:50:18 <gjanssens> warlord, fell: while jralls is in Japan, I'll be out of town as well from Thursday until the 18th. I have no idea whether I'll have decent network access where I'll be.
14:50:37 <gjanssens> So I may pop in from time to time if I can.
14:52:00 *** GabrieleV_ has joined #gnucash
14:52:08 *** GabrieleV has quit IRC
14:52:08 *** GabrieleV_ is now known as GabrieleV
14:52:12 <gjanssens> See you tomorrow
14:52:30 *** gjanssens is now known as gjanssens_
14:52:59 *** gjanssens_ has quit IRC
14:55:28 *** fabior has quit IRC
14:56:28 *** fabior has joined #gnucash
15:07:34 <Hinnerk> mathematician, that counts as .5 computers.
15:28:19 *** MechtiIde has quit IRC
15:28:48 *** mohammad has joined #gnucash
15:31:54 <Hinnerk> i think there are more tokens than transactions :)
15:32:02 <warlord> of course there are..
15:32:09 <warlord> each transaction will have multiple tokens
15:32:15 <Hinnerk> didn't think about it before.
15:32:28 <Hinnerk> talking about a cleaning up function:
15:32:44 <Hinnerk> I think it should allow to wipe token whose weight is 1.
15:32:55 <Hinnerk> these won't really help
15:34:04 <Hinnerk> Of course you can't clean these each time gnucash starts or closes.
15:34:25 <Hinnerk> should be triggered by clicking a button: "Prune Bayes"
15:34:44 <warlord> I think you only want to do that "after a while"...
15:34:50 <Hinnerk> of course
15:35:00 <warlord> You dont want to delete #1 if you're just starting out.
15:35:05 <Hinnerk> yes
15:35:15 <Hinnerk> but I see so many "dead" tokens
15:35:23 <Hinnerk> using dates or specific times.
15:36:38 <Hinnerk> and they are stuck in my file forever
15:36:40 <Hinnerk> and growing...
15:37:06 <Hinnerk> until i have a database of each minutes of a day (when I have withdrawn money at a versateller)
15:41:20 <Hinnerk> warlord: do you have a tool to check the integrity of the xml?
15:41:38 <warlord> nope. you can try 'xmllint'
15:58:23 <Hinnerk> coming to think about it: once oyu built up a database, you probably could just reverse the pruning: assume to always prune, unless you click a button.
15:59:09 <Hinnerk> in a way, I would be happy for Bayes not to learn anything new, but only use what is already there.
16:02:10 *** lmat_ has joined #gnucash
16:06:06 *** lmat has quit IRC
16:06:52 *** keret1 has joined #gnucash
16:11:13 *** StuM has joined #gnucash
16:11:17 *** keret has quit IRC
16:11:24 *** Krzysiek_K has left #gnucash
16:23:45 *** fell_ has joined #gnucash
16:23:46 *** gncbot sets mode: +o fell_
16:30:06 *** nomeata has joined #gnucash
16:33:06 *** fell has quit IRC
16:37:29 *** fabior has quit IRC
16:43:56 <Hinnerk> maybe the learning of bayes could also be improved.
16:44:05 <Hinnerk> I suspect it currently does the following:
16:44:16 <Hinnerk> each imported trx is split into tokens.
16:44:32 <Hinnerk> then, the algo is applied to calculate the other leg.
16:44:49 <Hinnerk> then manual corrections are done.
16:45:05 <Hinnerk> finally all new tokens are added and existing ones are "upgraded".
16:45:42 <Hinnerk> I think that for trx which have a high return value of the algo, it would be ok to not learn their new tokens.
16:48:25 <warlord> That does sound like the algorithm.
16:50:09 <Hinnerk> Need to switch computers, will be back in a moment.
16:50:15 *** Hinnerk has quit IRC
16:57:18 *** Hinnerk has joined #gnucash
16:57:25 <Hinnerk> rehi
17:11:44 *** fabior has joined #gnucash
17:12:52 *** himaxx has joined #gnucash
17:19:11 *** himaxx has quit IRC
17:32:37 *** himaxx has joined #gnucash
17:36:18 *** nomeata has quit IRC
17:52:23 *** jlo has joined #gnucash
18:07:02 *** fabior has quit IRC
18:42:03 <Hinnerk> after 2h of manual pruning bayes i reached token "A"
18:43:23 <kimmo2_> heh
18:43:31 <kimmo2_> what's token A?
18:46:49 <Hinnerk> just the single letter "A"
18:47:25 <Hinnerk> warlord: Actually even more important than a cleaning function would be to update the account names in bayes if you rename an account.
18:47:39 <Hinnerk> currently, information is lost and dead wood left in the file.
18:48:12 <kimmo2_> heh
19:20:02 *** jlo has quit IRC
19:21:42 *** jlo has joined #gnucash
19:44:58 *** rickoehn has quit IRC
19:53:18 *** jlo has quit IRC
19:54:21 <warlord> Hinnerk: I don't know why it uses names and not guids
19:55:19 *** aqua___ has quit IRC
20:02:40 *** ErKa has quit IRC
20:13:13 *** jlo has joined #gnucash
20:34:17 <Hinnerk> n8
20:34:20 *** Hinnerk has quit IRC
20:37:43 *** aqua___ has joined #gnucash
20:38:55 *** ErKa has joined #gnucash
21:06:16 *** lwells has joined #gnucash
21:07:05 *** GabrieleV_ has joined #gnucash
21:07:11 *** GabrieleV has quit IRC
21:07:12 *** GabrieleV_ is now known as GabrieleV
21:47:04 *** GabrieleV_ has joined #gnucash
21:47:18 *** GabrieleV has quit IRC
21:47:18 *** GabrieleV_ is now known as GabrieleV
22:17:04 *** GabrieleV_ has joined #gnucash
22:17:10 *** GabrieleV has quit IRC
22:17:11 *** GabrieleV_ is now known as GabrieleV
23:22:28 *** aqua_ has joined #gnucash
23:32:14 *** aqua___ has quit IRC
23:38:18 *** jlo has quit IRC
23:41:53 *** jimvideo has quit IRC
23:48:09 *** ErnstNeger has joined #gnucash
23:57:19 *** himaxx has joined #gnucash