2014-01-26 GnuCash IRC logs
00:31:21 *** ambimorph has quit IRC
01:52:57 *** ErKa has quit IRC
02:05:51 *** fell_ has joined #gnucash
02:05:51 *** gncbot sets mode: +o fell_
02:08:44 *** jmd has joined #gnucash
02:09:11 *** fell__ has joined #gnucash
02:09:12 *** gncbot sets mode: +o fell__
02:09:38 *** jmd has quit IRC
02:10:10 *** jmd has joined #gnucash
02:11:32 *** fell has quit IRC
02:14:45 *** fell__ has quit IRC
02:16:08 *** fell_ has quit IRC
02:16:33 *** jmd has joined #gnucash
02:18:35 *** Jimraehl1 has quit IRC
02:18:48 *** fell has joined #gnucash
02:18:48 *** gncbot sets mode: +o fell
02:21:36 *** fell_ has joined #gnucash
02:21:36 *** gncbot sets mode: +o fell_
02:21:36 *** fell has quit IRC
02:33:47 *** fell_ has quit IRC
02:49:42 *** fell_ has joined #gnucash
02:49:42 *** gncbot sets mode: +o fell_
02:54:51 *** fell__ has joined #gnucash
02:54:52 *** gncbot sets mode: +o fell__
02:55:10 *** fell_ has quit IRC
03:02:44 *** fell__ has quit IRC
03:42:54 *** TradeBorG113 has joined #gnucash
03:52:07 *** fell has joined #gnucash
03:52:07 *** gncbot sets mode: +o fell
03:59:46 *** Yawar has quit IRC
04:09:20 *** jralls has quit IRC
04:19:55 *** fell has quit IRC
04:48:38 *** jmd has quit IRC
04:48:40 *** jmd has joined #gnucash
05:02:03 *** aqua___ has quit IRC
05:02:10 *** gjanssens has joined #gnucash
05:02:11 *** gncbot sets mode: +o gjanssens
05:28:32 *** TradeBorG113 has quit IRC
05:30:21 *** TradeBorG113 has joined #gnucash
05:30:52 *** Jimraehl1 has joined #gnucash
05:44:14 *** aqua___ has joined #gnucash
07:41:00 <gjanssens> @tell warlord: the 2.6.1 buid failed
07:41:00 <gncbot> gjanssens: The operation succeeded.
07:44:21 <gjanssens> @tell warlord I removed the build dir and manually restarted the build. Let's see if this works. I may not be around to follow up
07:44:21 <gncbot> gjanssens: The operation succeeded.
07:44:59 <gjanssens> @tell jralls 2.6.1 build failed. I have manually restarted it, but may not be around to follow up. I have informed warlord as well
07:44:59 <gncbot> gjanssens: The operation succeeded.
08:09:03 <warlord> .
08:09:03 <gncbot> warlord: Sent 24 minutes ago: <gjanssens> I removed the build dir and manually restarted the build. Let's see if this works. I may not be around to follow up
08:09:19 <warlord> Why did it fail?
08:35:35 *** Jimraehl1 has left #gnucash
08:46:43 *** Jimraehl1 has joined #gnucash
09:04:21 *** ambimorph has joined #gnucash
09:52:25 *** aqua___ has quit IRC
10:03:52 *** aqua___ has joined #gnucash
10:35:42 *** fell has joined #gnucash
10:35:43 *** gncbot sets mode: +o fell
10:43:58 *** haclong has joined #gnucash
12:09:49 *** kpreid has quit IRC
12:10:46 *** kpreid has joined #gnucash
12:25:02 <gjanssens> warlord: It can't find libpthread-2.dll
12:25:10 <gjanssens> I still have to figure out why
12:30:15 <warlord> Huh -- isn't that part of the install? Or is it out of space?
12:33:48 <warlord> I dont understand why it would be in the trunk area but not in a tag build.
13:20:26 *** jralls has joined #gnucash
13:20:27 *** gncbot sets mode: +o jralls
13:30:25 *** ErKa has joined #gnucash
13:36:50 <gjanssens> warlord: it's indeed a diskspace issue
13:37:57 <gjanssens> With the restarted build, there currently is about 250 Mb free and it's just started to build gmp
13:38:10 <gjanssens> I think I'll remove the 2.4 weekly build directory
13:38:19 <gjanssens> That should free sufficient diskspace
13:38:36 <gjanssens> And we have no plans anymore to make changes in that branch
13:39:37 <jralls> Ah, answered my question without having to ask.
13:39:37 <gncbot> jralls: Sent 5 hours and 54 minutes ago: <gjanssens> 2.6.1 build failed. I have manually restarted it, but may not be around to follow up. I have informed warlord as well
13:39:50 <gjanssens> Or should I rather throw away the build directories of the old release builds ?
13:40:34 <jralls> Certainly the 2.4 weekly can go. No need for those any more.
13:41:30 <jralls> What's the point of maintaining old release build directories? After the build is successful, we're done with the build directory aren't we?
13:42:05 <gjanssens> I guess so
13:42:16 <gjanssens> But I notice they are not removed automatically
13:42:28 <jralls> Easy to fix...
13:42:28 <gjanssens> For post mortem analysis perhaps ?
13:43:15 <gjanssens> I was thinking to keep the 2.4 weekly directory around until we set up the 2.6 equivalent
13:43:36 *** jmd has quit IRC
13:43:44 <gjanssens> I don't remember if we had any special configuration tricks for it or not
13:43:48 <jralls> Like I said, once the release completes successfully, we're done with it. If it doesn't complete successfully then a post-mortem is needed; but once the cause of the failure is determined and fixed, then the build directory gets over-written by the restart.
13:44:24 <gjanssens> True, so I'll drop all the release directories that are still around
13:46:02 <jralls> What's in the 2.4 weekly directory that would be useful? I guess if you renamed it to 2.6 that it might save some time on the first 2.6 build, but it will also be cluttered with cruft from earlier versions of dependencies and old dependencies that 2.6 doesn't need.
13:46:47 <gjanssens> It can't be reused directly. 2.4 was an svn driven build, 2.6 is git driven
13:46:58 <gjanssens> But I like to keep a reference around
13:47:30 <gjanssens> Looking at an actual setup often works like reading a manual for me
13:47:48 <gjanssens> We'll have plenty of space available when I remove the release directories
13:48:06 <jralls> Wait, are we talking about the script or the build directory?
13:48:25 <gjanssens> Each of these directories contains a copy of the downloads directory (with all tarballs and such)
13:48:35 <gjanssens> I was really talking about the directory
13:48:48 <gjanssens> I will drop the build line from the script alright
13:48:53 <gjanssens> So the build will not run anymore
13:49:19 *** Yawar has joined #gnucash
13:49:20 <jralls> >copy of the downloads: That's dumb. All builds should use a common download directory.
13:49:30 <gjanssens> Anybody know by the way the reason we copy the download directory ?
13:49:47 <gjanssens> I was about to propose a common directory as well
13:50:54 <jralls> My guess would be that it was marginally easier to script, but Derek doesn't seem to be here to defend himself.
13:51:40 <jralls> But it's probably not Derek anyway; those scripts are all part of GnuCash, right?
13:53:33 <gjanssens> Yes
13:57:32 <gjanssens> Got that fixed in r23774
13:59:25 <gjanssens> So far I've only removed 3 release build dirs -> saves 4.5Gb already...
13:59:46 <gjanssens> The others will follow next
14:01:39 <warlord> gjanssens: I usually go in every once in a while and clean up a bunch of tag-build dirs and clean up old output.
14:02:45 <gjanssens> What do you mean with "old output" ?
14:02:58 <gjanssens> I'm removing the old tag-build dirs now
14:03:19 <warlord> gjanssens: I copied the build dir so that each tag was unique but windows doesn't have an "LN" command to hard-link
14:03:33 <warlord> jralls: No, it was me who did that.
14:04:04 <warlord> old output == old setup.exe and build logs.
14:04:14 <warlord> (from weekly and daily builds)
14:05:16 <jralls> Can't you make Scheduler do that so that you don't have to remember?
14:05:33 <Yawar> warlord: btw, windows has the 'mklink' command which is equivalent to 'ln'
14:06:35 <Yawar> or at least, ln -s
14:06:50 <warlord> jralls: I was using human choice to decide what to erase, rather than some computational algorithm. For example, on 'code' I clear out old builds periodically but I keep the "stable" versions, for example.
14:07:00 <warlord> Yawar: I suspect that at the time I didn't know about that command ;)
14:07:09 <warlord> (this was written a long time ago)
14:07:16 <warlord> ... like when 2.4 first came out.
14:07:17 <warlord> and on XP
14:07:28 <Yawar> ah, xp is a different matter :-)
14:08:44 <warlord> Yeah, the build server is XP
14:11:44 <Yawar> this may be a bad segue, but i'm wondering if anyone has been working on putting unit tests into the gnucash scheme reports?
14:11:48 <jralls> Which brings up an interesting point. Geert, have you tested your new build scripts on XP?
14:12:23 <gjanssens> My windows test box is window XP as well, so yes
14:12:40 <jralls> OK.
14:12:59 <gjanssens> Yawar, recently a few unit tests were added to the reports
14:13:27 <gjanssens> They came with some performance improvements
14:16:30 <Yawar> geert, are those the commits from peter broadbery? if so i will take a look
14:16:57 <gjanssens> Yes those are the ones
14:17:17 <gjanssens> Actually I committed them, but they are from peter broadberry
14:17:38 <gjanssens> I think I mentioned his name each time in the commit messages
14:18:42 <gjanssens> jralls warlord while on the topic of windows builds
14:19:10 <Yawar> got it, thanks...
14:19:15 <gjanssens> I'm considering to rip the packaging directory from the gnucash repository and create an independent one
14:19:43 <gjanssens> similar to the gnucash_on_osx independent repo
14:20:55 <warlord> I would rather see the 'gnucash_on_osx' get migrated into gnucash.
14:21:21 <gjanssens> No, they should be separate repos under the gnucash umbrella
14:21:37 <gjanssens> all available on code or github/gnucash
14:22:34 <gjanssens> but 90% of the people that would clone the gnucash repository don't need the windows/os x specific build scripts
14:22:52 <gjanssens> Having them all in one repo would only increase the size of the data to download
14:23:35 <gjanssens> Also the build scripts don't follow the release cadence of gnucash
14:24:05 <gjanssens> And that's not necessary
14:24:44 <gjanssens> You can build the same gnucash release on different distro versions as well
14:25:58 <Yawar> imho with git when you get a chance to make your repo smaller by splitting into multiple, you usually take it....
14:26:15 <gjanssens> That's how I see it as well
14:26:17 <Yawar> ... because unlike svn git doesn't give you the option to check out a few files only
14:26:55 <gjanssens> A similar thing should happen once on the gnucash-docs repo imo
14:27:28 <gjanssens> Correction: s/gnucash-docs/gnucash-htdocs/
14:28:05 <gjanssens> That repo currently has the phtml source for the website *and* a copy of all the released docs
14:28:28 <gjanssens> It would make more sense to make a separate repo for the docs
14:28:38 <warlord> Except there are often lock-steps between the packaging rules and the main code, e.g. when you change the minimum dependencies.
14:29:36 <warlord> and it's not like the the packaging code is all that large.
14:29:41 <gjanssens> The web pages get updated much more than the docs are released
14:30:05 <gjanssens> And I had fell complain about this before (rightly so IMO)
14:30:39 <gjanssens> The lock-step issue exists on distos as well right ?
14:30:44 <warlord> We already have a separate repo for the docs
14:30:57 <warlord> "distos"?
14:31:00 <jralls> I'm not real wild about having the document binaries in VC at all.
14:31:02 <gjanssens> Yes, but not for the *built* versions of the docs
14:31:39 <gjanssens> I know. This has grown so because that's the only way to get them on www.gnucash.org
14:31:44 <warlord> Except for the website, we dont have built-docs in VC anywhere.
14:31:52 <warlord> right..
14:31:54 <jralls> They're there because it makes it convenient to push them to the web server without Linas having to worry about access rights.
14:31:59 <warlord> Unless we wanted to point all docs users at Code.
14:32:22 <jralls> Or SF.
14:32:25 <gjanssens> Which is fine for me as well
14:32:40 <gjanssens> Can you put html on sf ?
14:32:41 <warlord> How does SF help us? they can't do a daily build of gnucash-docs?
14:32:48 <fell> where is the problem to push htdocs and nightlydocs separate?
14:33:00 <gjanssens> nightlydocs are on code
14:33:07 <gjanssens> htdocs is on www
14:33:18 <warlord> and there is no "nightly docs" repo
14:33:21 <gjanssens> code is managed by warlord, www is managed by linas
14:33:32 <jralls> SF can host the build product for us. I was thinking more of the released docs than the nightly builds.
14:33:34 <fell> I know.
14:34:10 <jralls> Yes, you can run a whole web site on SF.
14:34:15 <fell> But can we not split htdocs in pure htdocs and ready made docs?
14:34:40 <warlord> fell: I see no reason to do that; the built-docs dont get updated frequently.
14:35:04 <jralls> They do now. I'm releasing new docs with every GC release.
14:35:18 <fell> Once I wanted to fix a typo and had to download all the ready made docs and my time window was over.
14:35:19 <warlord> then again I'm pulling off the LAN so I don't see much network latency on a clone or pull ;)
14:35:30 <gjanssens> And in addition it makes the htdocs repo grow significantly
14:35:35 <warlord> True
14:35:48 <warlord> We could work with Linas to set up a separate pull
14:35:48 <gjanssens> Which is not useful for people only wanting to update a webpage
14:36:03 *** ErKa has quit IRC
14:36:26 <warlord> Or we could do it as a submodule in htdocs (which would still require a second pull on www)
14:37:41 <gjanssens> Those are assuming a separate repo
14:37:49 <Yawar> if you're talking about a git submodule, i would advise avoiding those
14:38:10 <Yawar> i don't think any of us here has sufficient git prowess to handle those
14:38:19 <jralls> Submodules are a pain. A separate repo would work as well, and it would shrink a lot if only the HTML was in it. The PDF, mobi, and ebook versions could go to SF with a static link on the website.
14:39:43 <jralls> Yeah, and Derek used to think that none of us had commit on Gnome. I've used submodules. They work, but they're a pain, even after the improvements in git 1.7.
14:39:46 <gjanssens> If you put those on sf, would that mean there would always be the annoying "Download will start in a few seconds" commercial page ?
14:40:07 <jralls> Yeah, it's how they pay their bills.
14:40:57 <jralls> Downloads are expensive. That's why Github and Google Code stopped supporting them.
14:42:23 <warlord> Yes, I realize submodule requires a second repo -- isn't that what y'all are asking for? My issue is that git doesn't really let you do a clone inside another cloned dir.. So e.g. /var/www/html is a clone of htdocs, we couldn't clone built-docs into /var/www/html/docs/
14:42:57 <warlord> this is why I think a submodule would be good; putting the built-docs repo into the 'docs' subdir of the 'htdocs' repo
14:44:15 <gjanssens> It would probably be much easier if Linas set up an httpd alias
14:44:22 <gjanssens> And keep two independent repos
14:44:43 <gjanssens> On the filesystem those will be two directories
14:44:59 <gjanssens> Via the web, docs would appear as a subdirectory of www
14:45:12 <gjanssens> That's a one line change in the apache config files
14:45:41 <gjanssens> Re the price of downloads
14:46:11 <gjanssens> So far Linas doesn't seem to mind the bandwidth we use
14:46:37 <gjanssens> So we could keep all files on www and avoid the sf commercials
14:47:05 <gjanssens> I have no idea of the bandwidth we use by the way
14:47:26 <gjanssens> So perhaps code could even be considered as a download location for the binary formats
14:47:34 <gjanssens> But that's warlord's call
14:48:03 <jralls> For built docs, probably not that much. You're not proposing hosting the gnucash binaries on code, are you?
14:48:29 <gjanssens> Nope
14:53:26 <gjanssens> jrall fyi it seems the build is working now
14:53:37 <gjanssens> It's currently building gmp
14:53:53 <gjanssens> It will take quite a while still though
14:54:41 <jralls> Good. Fingers crossed (for luck, not dissembling).
14:54:49 <gjanssens> And regarding automatically deleting the tag builds after a successful build
14:54:54 * jralls is working on the news article.
14:55:04 <gjanssens> There's a small quirk we have to fix then
14:55:17 <gjanssens> Each tag build installs its own version of git
14:55:40 <gjanssens> And the last install apparently can't be deleted just like that
14:55:56 <fell> jralls: you should mention the docs are released simultanous.
14:55:58 <gjanssens> The git application is running permanently
14:56:18 <jralls> Frank: Of course. One article or two?
14:56:39 <gjanssens> I guess to cater for the context menu items in the windows explorer
14:56:58 <fell> Doesn't matter if it is obvisiuos.
14:57:14 * jralls is leaning towards one, to emphasize that they go together.
14:57:26 <gjanssens> That problem will probably fix itself if we make the packaging dir an independent repo
14:57:44 <gjanssens> In that case it's not gnucash anymore that installs git
14:58:14 <gjanssens> It's not now either strictly speaking
14:58:20 <gjanssens> Oh well...
14:58:33 <jralls> Surely there's a git package in MinGW nowadays.
14:58:47 <gjanssens> You'd be surprised :(
14:58:50 <jralls> So that we don't have to get the standalone one.
14:58:55 <gjanssens> There is not
14:59:07 <jralls> Sigh.
14:59:12 <gjanssens> Git on windows has cloned mingw and made some incompatible changes
14:59:26 <gjanssens> Mingw has had their long toes stepped upon
14:59:32 <gjanssens> And refuse to integrate git
14:59:41 <jralls> That was rude of the git folks.
14:59:49 <gjanssens> It is
15:00:09 <jralls> OTOH, it's dumb of the MinGW folks.
15:00:09 <warlord> eh, they feel they are the center of the universe.
15:00:09 <gjanssens> But I'm not sure they had another choice
15:00:33 <warlord> sorta like the libdbi people ignoring issues with 32-bit time_t
15:00:35 <gjanssens> I don't know the details
15:01:34 <gjanssens> But the mingw mailing list has plenty of messages telling people "this is not mingw's issue, we don't care"
15:02:05 <gjanssens> That kind of invites people to do things on their own eventually
15:02:20 <gjanssens> (64-bit mingw anyone ?)
15:02:38 <jralls> It wouldn't surprise me if the git folks offered their changes upstream and MinGW in the person of Keith Marshall told them to buzz off.
15:02:59 <gjanssens> Indeed
15:03:14 <jralls> I've seen plenty of examples of him doing that.
15:06:43 <warlord> back to www.. *IF* we could get Linas to set up the apache alias, then yes, /var/www/html and /var/www/gnc-docs would work fine.
15:07:10 <warlord> ... and then we would need multiple push/pulls
15:08:55 <warlord> I'm happy to set that up on the code side; we would need him to set it up on www
15:16:25 *** TradeBorG113 has quit IRC
15:19:39 <gjanssens> True
15:19:49 <gjanssens> I do have access to www and can prepare most of it
15:20:04 <gjanssens> But the system level changes are for linas
15:20:41 <gjanssens> So there's first an if and the a when
15:21:11 <gjanssens> I don't think any of this can be done in time for your migration of code
15:21:31 <gjanssens> That's not really necessary either
15:22:17 <warlord> No, it's not necessary.. it can happen on its own time frame.
15:22:18 <gjanssens> It can be done at any time, but we should inform the community about it
15:22:22 <warlord> linas__: are you around?
15:39:12 <warlord> we can always send linas email to ask. I'm sure he'd be okay.
15:39:30 <warlord> Anyways, gotta run. BIAW
15:51:49 *** aqua___ has quit IRC
15:52:43 *** haclong has quit IRC
16:00:25 *** Jimraehl1 has quit IRC
16:03:37 *** ErKa has joined #gnucash
16:24:04 *** ErKa has quit IRC
16:27:37 *** aqua___ has joined #gnucash
16:37:24 *** Yawar has quit IRC
16:51:17 *** haclong has joined #gnucash
17:11:43 <gjanssens> jralls: the build is now in isocodes
17:12:12 <gjanssens> ...and I am ready for bed :)
17:12:19 <gjanssens> Good luck with the release
17:12:24 <gjanssens> See you later
17:13:58 <jralls> G'night, Geert.
17:14:43 *** gjanssens has quit IRC
17:53:35 *** wilbert has joined #gnucash
17:54:49 *** wilbert has left #gnucash
18:17:12 *** ErKa has joined #gnucash
19:16:15 *** ErKa has quit IRC
19:16:21 *** haclong has quit IRC
19:28:31 *** haclong has joined #gnucash
20:08:35 *** haclong has quit IRC
21:45:08 *** ambimorph has quit IRC
22:39:05 <warlord> Hmm, I still dont see a log for 2.6.1 on code. too tired to login to the builder.
23:25:30 *** jralls has quit IRC
23:26:59 *** jralls has joined #gnucash
23:26:59 *** gncbot sets mode: +o jralls
23:35:52 *** Jimraehl1 has joined #gnucash