2007-10-13 GnuCash IRC logs

00:01:21 *** dbr has quit IRC
01:20:07 *** _lasindi has joined #gnucash
01:20:08 *** lasindi_ has quit IRC
02:06:19 *** Jaran has quit IRC
03:41:56 *** Zoolooc has joined #gnucash
04:40:37 *** Jaran has joined #gnucash
04:54:06 *** basanta has joined #gnucash
05:36:42 *** basanta has joined #gnucash
05:57:35 *** pdc has quit IRC
06:09:46 *** roxy_ has joined #gnucash
06:11:17 *** roxy_ has left #gnucash
06:36:35 *** pdc has joined #gnucash
06:55:05 *** basanta has quit IRC
08:01:24 *** Rol1 has quit IRC
08:10:58 *** andi5 has joined #gnucash
08:10:58 *** gncbot sets mode: +o andi5
08:11:27 *** pecisk has joined #gnucash
08:32:42 *** warlord-afk is now known as warlord
09:00:59 *** eauxnguyen has quit IRC
09:05:33 *** eauxnguyen has joined #gnucash
10:54:40 *** andi5 has left #gnucash
11:13:48 *** kielein has joined #gnucash
11:27:15 *** cstim has joined #gnucash
11:27:15 *** gncbot sets mode: +o cstim
11:27:34 <cstim> @tell andi5 Thanks for the 2.2 branching
11:27:34 <gncbot> cstim: The operation succeeded.
11:29:01 <warlord> Yes.. yay!
11:30:29 <cstim> warlord: hi. long time no see.
11:30:52 <warlord> Indeed! How are you?
11:31:01 <warlord> (I leave for the airport in about 20-30 minutes)
11:32:17 <cstim> I'm fine. Still trying to finish the phd thesis but being very bored by that.
11:32:41 <cstim> Maia's doing very well. She really speaks a lot by now.
11:33:08 <warlord> Nice!
11:33:17 <warlord> My eldest nephew still is only speaking a little.
11:33:43 <cstim> and (any parents around here) we're excited that she managed to use the potty already
11:34:58 *** andi5 has joined #gnucash
11:34:58 *** gncbot sets mode: +o andi5
11:35:04 <andi5> hi cstim :-)
11:35:49 <andi5> cstim: btw, you have been to fast for me... i had the code ready for win32, but was too tired to test-compile it on linux again (-#include <sys/mman.h)
11:36:04 <cstim> andi5: heh
11:36:14 * cstim needed some distraction from the thesis...
11:38:12 *** wizkid239 has quit IRC
11:39:10 <cstim> Back in the old days, when one latex run took considerable amount of time, you were better off to process only a subset of the full text.
11:39:30 *** wizkid238 has joined #gnucash
11:39:31 <cstim> which can be achieved by \include and \includeonly quite nicely.
11:39:38 <andi5> yep, i know
11:39:51 <andi5> maybe mine was too short (only 50 pages)
11:39:58 <cstim> but nowadays I think nobody uses that anymore. Hence, one big file works just as well as many files.
11:40:25 <cstim> I'm at 125 right now and I try to arrive at 150.
11:40:46 <andi5> you are cheating with figures, right? ;-)
11:40:56 <cstim> I think with more than 200 pages you would still use \includeonly today, but not with smaller texts.
11:41:28 <cstim> cheating? figures are vital for the understanding! :-) Coincidentally they also consume a lot of page space :-)))
11:42:04 <andi5> just because they are so important! :) .... let us use 9 pages for notes after each figure :
11:42:44 <cstim> andi5: have you heard about race conditions with Windows sockets and their send() function; in aqbanking they seem to be avoided only by a Sleep(300) before send(): http://bugzilla.gnome.org/show_bug.cgi?id=439654#c20
11:43:38 <andi5> i have heard about sleeps necessary after wget... and i really wonder what.... windows sometimes is
11:44:00 <cstim> wget?
11:44:15 <cstim> any reference for further reading?
11:44:39 <andi5> http://svn.gnucash.org/trac/changeset/16524
11:45:26 <cstim> oh, *that* wget :-)
11:45:55 <cstim> Have you ever programmed with sockets before? maybe even on windows? /me didn't
11:46:04 <andi5> btw ... are there any plans how we handle backport requests this time?
11:46:12 <andi5> on windows? no...
11:47:18 <cstim> you mean http://wiki.gnucash.org/wiki/Development_Process ? IMHO we can go back into that mode of operation. OTOH it's unclear whether it really helps right now...
11:48:55 <warlord> Okay, I gotta logout soon..
11:49:12 <cstim> warlord: have a good trip
11:49:15 <warlord> cstim: I think it might still be useful to audit patches for backport to 2.2
11:50:01 <andi5> hm... to throw in (repeat) another question: what about the other branches? ... (no, i do not mean r-r)
11:51:12 <cstim> which ones?
11:51:59 <andi5> hm... i was under the impression that there is at least one other...
11:52:13 <cstim> the other summer of code results?
11:52:21 <andi5> yep
11:52:27 <cstim> there was dogtail and this can surely be merged
11:52:34 <andi5> ah, right
11:52:45 <andi5> that slipped my scanning
11:52:47 <warlord> You should talk to jsled about dogtail.
11:52:58 <warlord> the only other GSoC branch is r-r.
11:53:39 <cstim> and qif-rewrite but that project failed.
11:54:00 <warlord> yes.
11:54:11 <warlord> okay, gotta run
11:54:12 <warlord> bye
11:54:22 <andi5> bye warlord
11:54:27 <cstim> bye
11:54:54 <warlord> bye.
11:55:07 <warlord> talk to you from california.
11:55:27 *** warlord has quit IRC
11:55:27 <cstim> I thought the dogtail branch only added new subdirectories (src/test-dogtail) and doesn't interfere with any of our work.
11:55:45 <andi5> yep, i hope so
11:55:52 <cstim> but I don't have a repo available here to check the actual diff.
11:56:05 <andi5> i could make one
11:56:08 <cstim> andi5: you still run svn+svk?
11:56:17 <andi5> i use to use git-svn
11:56:25 <cstim> you use to use? :-)
11:56:31 <andi5> i use... whatever
11:56:55 <andi5> maybe "am used to use"? ... hey
11:57:25 <cstim> I have git-svn but I downloaded only trunk and only the last year or so. But I don't know how to manage branches/2.2 in the same git-svn repo. Any ideas?
11:57:59 <andi5> simply add a line to .git/config
11:58:25 <andi5> iirc this worked pretty well with recent git-svn versions
11:58:39 <andi5> let us try it...
11:58:48 <cstim> which line into the config?
11:59:27 <andi5> in the setion "svn-remote svn" (i have only one section svn-remote) , add a fetch line
11:59:55 <cstim> currently I have these two lines:
12:00:06 <cstim> url = svn+ssh://cstim@svn.gnucash.org/repo/gnucash/trunk
12:00:10 <cstim> fetch = :refs/remotes/git-svn
12:00:32 <andi5> http://pastebin.ca/raw/735379
12:00:58 <cstim> hey, that looks *good*
12:01:09 <andi5> this is with git-svn version 1.5.2.5
12:02:02 <cstim> there's a typo with the remotes/2.2 line
12:02:11 <andi5> eek,
12:02:23 <andi5> thanks
12:02:51 <chris> hey andi5, cstim
12:02:59 <andi5> hi chris
12:03:06 <chris> http://pastebin.com/m3009922f
12:03:23 <andi5> yeah, that is for all branches, right?
12:03:27 <chris> yeah
12:04:25 <chris> andi5: you should check out 1.5.3
12:04:29 <chris> or 1.5.4
12:04:35 <andi5> why that?
12:04:51 <chris> git-rebase -i, for one.
12:05:02 <cstim> chris: you should put this on http://wiki.gnucash.org/wiki/git
12:05:21 <chris> git-svn rebase for another, at least I think that was new-in-1.5.3
12:05:46 <chris> cstim: which, the config?
12:05:50 <andi5> chris: and please add a warning about its initial download size :-)
12:05:55 <cstim> chris: yes
12:07:04 <chris> I don't know if it's such a good idea to encourage casual developers to do what I did, though.
12:07:22 <cstim> I think the selected branches are enough for me.
12:07:37 <chris> I tend to do a lot of archeology.
12:07:50 <andi5> long live git-gc :-)
12:07:55 <cstim> chris, the archeologist :-)
12:08:24 <chris> git makes it so easy.
12:09:07 <andi5> chris: have you used git-rebase -i a lot?
12:09:24 <chris> so, I don't know what to put on the wiki. I think my config is basically the default from git-svn init.
12:09:30 <chris> andi5: *constantly*
12:09:48 <andi5> Error from SVN, (210004): Fehlerhafte Netzwerkdaten: Fehlerhafte Netzwerkdaten
12:10:18 <andi5> hm... worked this time :-)
12:11:10 <chris> git-rebase -i is probably my favorite feature.
12:11:52 <andi5> what do you use it for? do you make normal rebases, just without one single bad commit?
12:12:08 <andi5> or do you smash several commits together?
12:12:09 <chris> although, {un,}staging single hunks from git-gui is pretty darn awesome, too.
12:14:21 <chris> andi5: all of the above actually. I always work on a private branch, and I git rebase -i BRANCH-BASE to basically rewrite/reorder/edit/squash/amend whatever the entire branch.
12:15:52 <chris> often, I'm committing fixes for my previous commits, so after a while, I squash the fixes into the original commits.
12:16:05 <andi5> i think i would be even more happy if git-svn supported all svn properties, most importantly svn:ignore and svn:externals
12:17:13 <andi5> how do you do that?
12:17:55 <chris> svn:ignore is NoBigDeal: git svn show-ignore >> $GIT_DIR/info/exclude
12:18:09 <andi5> and the other direction?
12:18:43 <chris> svn:externals are trickier, and there's no general-solution, although people have talked about using git-submodules.
12:19:03 <chris> andi5: nothing in the other direction. :(
12:19:08 <andi5> oh, git-submodules is ready?
12:19:57 <chris> it's released, and apparently, people use it.
12:20:28 <chris> I'm not sure it fits everybody's ideal workflow, though.
12:20:57 * chris hasn't used git-submodule
12:21:37 <andi5> suppose you have some branch "chris" with a commit A you want to change... what do you do? i am interested in your workflow :)
12:21:38 <chris> andi5: how do I do what?
12:21:53 <andi5> or commits A..B
12:21:56 *** nbinont-afk has joined #gnucash
12:22:55 <andi5> (i suppose this depends on the fact that nobody is pulling from your branch)
12:23:47 <chris> yes, git-rebase -i A
12:24:35 <chris> it depends on what you want to change. If it's just the commit message, I mark the commit with 'edit'.
12:25:36 <chris> You can also mark an intermediate commit as edit, and then do whatever you want: add-more commits, amend, whatever, and then git-rebase --continue.
12:26:05 <andi5> "mark with edit" ... is that something i do with git-gui?
12:26:15 <andi5> i have got the feeling i know NOTHING about git ;-)
12:26:43 <chris> although, I don't do that often. I prefer to just add new commits before the rebase, and then squash them with other commits during the rebase.
12:27:12 <chris> andi5: It certain sounds liek you've never used git-rebase -i :)
12:27:20 <andi5> hehe :)
12:35:59 <chris> (sorry, network problems here)
12:36:40 <chris> git-rebase -i brings up an editor to allow you to reorder/mark for edit/whatever.
12:37:45 <cstim> chris: when I change my .git/config as discussed above, it seems git-svn will not fetch the branches in addition to what it used to fetch so far.
12:38:32 *** nbinont-afk has quit IRC
12:38:39 <chris> cstim: I think it only affects new revisions.
12:39:21 <chris> I actually started from the beginning like that.
12:39:39 <cstim> chris: but that will cause a humongous download...
12:41:26 <chris> It did. once upon a time, long ago.
12:42:30 <cstim> chris: do you know how I can remove some fetched commits on the remote branches so that I can have them fetched again?
12:45:26 <andi5> cstim: remove the index.db and reset --hard the remote branches head
12:45:40 <cstim> really?
12:45:48 <andi5> btw, i issued "git svn fetch" and it worked (took a while)
12:45:48 <cstim> removing the index.db, that is.
12:46:06 <andi5> or .rev_db
12:46:37 <andi5> the .rev_db maps svn revisions (line number -1, i think) to commits
12:47:43 <chris> yeah, I guess that would work, but I think that's pretty much the same as creating a new [svn-remote].
12:48:06 <chris> In terms of efficiency , I mean.
12:48:11 <andi5> no, the .rev_db's are stored per fetch
12:48:18 <andi5> the recreation of that file is FAST
12:48:28 <chris> really?
12:48:31 <cstim> there is no .rev_db here
12:48:44 <andi5> in .git/svn/$fetch
12:48:53 <cstim> git --version says 1.5.3
12:49:03 <andi5> oh, maybe they changed that again?
12:49:10 <chris> I have the same.
12:50:48 <chris> cstim: are you sure you have no .git/svn/*/.rev* ?
12:52:14 <chris> BTW, my .git is 264M
12:53:34 <andi5> http://pastebin.ca/raw/735435
12:56:36 <cstim> chris: oh, there is .rev_db.somethingverylong
12:56:38 <cstim> see ya
12:56:42 *** cstim is now known as cstim_dinner
12:57:03 <andi5> i guess i will have to fetch something to eat
12:57:27 * andi5 andi5@store
12:58:19 *** andi5 is now known as andi5|store
12:58:24 <chris> andi5: you know about git-diff --stat, right?
12:58:33 *** cstim_dinner has quit IRC
12:58:57 <andi5|store> yes, but yesterday i needed the patch, ...
12:59:32 <andi5|store> at least that is how i do git-svn merging right now....
13:00:16 <andi5|store> (i rebase prior to that)
13:00:55 <chris> ah, I see.
13:01:45 <andi5|store> well, it sucks and i know that there are better ways, but csv-import was easy to merge, so...
13:14:41 *** lasindi_ has joined #gnucash
13:14:41 *** _lasindi has quit IRC
13:48:36 *** andi5|store has quit IRC
14:01:29 *** Rolf has joined #gnucash
14:20:11 *** andi5 has joined #gnucash
14:20:11 *** gncbot sets mode: +o andi5
14:30:50 *** McLANE has joined #gnucash
14:35:48 *** jeez has joined #gnucash
14:41:39 <jeez> Hi. Quick question: Since there is no OS X binary and I don't let Fink on my system, where can I get a complete dependency list?
14:43:43 *** tjanssen has joined #gnucash
14:43:54 <andi5> jeez: do you know http://wiki.gnucash.org/wiki/MacOSXInstallation?
14:46:46 *** McLANE has quit IRC
14:50:24 *** McLANE has joined #gnucash
14:50:29 *** Rol1 has joined #gnucash
14:50:47 <jeez> Yeah. It basically says "use Fink". However, I think that Fink is borderline unusable, combining Debian's lack of flexibility with Gentoo's compile times (as binary packages and "unstable" mode don't mix). Since Portage/OS X doesn't have Gnucash yet I'd like to see which dependencies it can build; the rest would be added in with custom ebuilds.
14:52:22 <jeez> (Yes, it also gives the option of using MacPorts, but MacPorts is a nightmare and I want to avoid having multiple package managers under one OS. It tends to be decidedly unpretty.)
14:52:25 <andi5> jeez: so do you know http://svn.gnucash.org/trac/browser/gnucash/trunk/README.dependencies ? (i hope it is state-of-the-art)
14:53:04 <jeez> Ah. That does look useful, indeed. Thanks.
14:53:31 <andi5> you are welcome
14:58:00 *** Rolf has quit IRC
15:12:28 *** jeez has quit IRC
15:37:03 *** kielein has quit IRC
15:43:37 *** Zoolooc_ has joined #gnucash
15:46:57 *** Zoolooc__ has joined #gnucash
15:52:08 *** Zoolooc has quit IRC
15:52:38 *** Zoolooc has joined #gnucash
15:54:38 *** Zoolooc_ has quit IRC
15:55:14 *** Zoolooc_ has joined #gnucash
16:00:08 *** Zoolooc__ has quit IRC
16:00:17 *** Zoolooc__ has joined #gnucash
16:02:14 *** McLANE has quit IRC
16:03:38 *** Zoolooc has quit IRC
16:03:44 *** Zoolooc has joined #gnucash
16:07:38 *** Zoolooc_ has quit IRC
16:10:19 *** Zoolooc__ has quit IRC
16:18:56 *** tim_abell_ has joined #gnucash
16:41:18 *** dorky has quit IRC
17:15:12 *** Def has joined #gnucash
18:30:24 <andi5> chris: ping
18:57:36 *** tim_abell__ has joined #gnucash
18:57:38 *** tim_abell__ is now known as tim_abell
19:03:26 *** tim_abell_ has quit IRC
19:06:52 *** pecisk has quit IRC
19:41:30 <chris> andi5: pong (slow)
19:41:41 <andi5> yeah ;-)
19:44:59 <andi5> regarding r-r, i wonder what the wanted input behavior is... 1) i suppose <enter> should become "commit and goto next" again, right? 2) what about the activation of cells? should there always be an active cell iff the anchor account of the register is not a placeholder? 3) do you think the blank transaction could be sorted as well, so that entering data in the past does not involve a lot of scrolling?
19:46:01 <andi5> 4) are there movement keybindings in the date-cell-rendered and are they customizable/i18n'ed?
19:48:40 <andi5> 5) if the selection is on one of several entries with the same date, it is always showed at the bottom... i think you have seen it before as well... i think this is due to the reparentization of the blank split... i still have to think about how to resolve that... any ideas?
19:49:10 * andi5 thinks
19:52:11 <chris> I'll have to respond to others after I grout my bathroom tile, or I'll have no place to shower, but: 1) yes, 4) I think so.
19:52:53 <andi5> have fun :-)
20:38:12 *** andi5 has left #gnucash
21:22:44 *** IanL has joined #gnucash
22:02:19 *** Rol2 has joined #gnucash
22:03:24 <chris> 2) There should always be an "active" cell, meaning a cell with the focus, even with placeholders, as it is the cursor for navigation. However, that cell should only become editable upon the first edit-key, and only if the anchor is editable, of course.
22:05:20 <chris> 3) I think the blank transaction should be sorted, but I'm not sure I understand what that implies for scrolling. I don't think its order should change based on uncommited fields. I think its order should only be based on the default date.
22:06:50 <chris> 5) I think you're talking about the GtkTreeSortable-order-among-identical-sort-keys-depends-on-event-history-"bug".
22:08:11 <chris> 5) only solution would be to always add an additional sort-field with some stable, but possibly arbitrary, ordering.
22:08:20 <chris> 5) s/only/one/
22:10:00 *** Rol1 has quit IRC
22:10:06 <chris> @tell andi5 http://lists.gnucash.org/logs/2007/10/2007-10-13.html#T22:03:24
22:10:06 <gncbot> chris: The operation succeeded.
22:16:14 *** benoitg has joined #gnucash
22:17:51 *** Zoolooc_ has joined #gnucash
22:27:10 *** Zoolooc has quit IRC
23:19:11 *** cparker has quit IRC
23:33:52 *** tjanssen has quit IRC