[00:00] bug reported, thanks. https://bugs.launchpad.net/bzr/+bug/637644 [00:00] Launchpad bug 637644 in Bazaar "Repository corruption, index file truncated by OS file system (affected: 1, heat: 6)" [Undecided,New] [00:01] hi folks, i branch a repo from lp in my VPS now i want to get it to my laptop i get this error: bzr: ERROR: Unable to connect to 173.230.141.58; [Errno 111] Connection refused [00:01] trying: bzr clone ftp://XXX.XXX.XXX.XXX/home/myuser/openerp/stable/addons [00:37] ah people... [00:38] like free online support has to be instant [00:43] hello ddaa [00:43] hello poolie, wassup? [00:43] mm i was going to answer but he just left [00:44] ah, things are pretty good here [00:44] so, my bet was "missing username from URL" :-) [00:44] just going to close all my other windows in a bit and look at this Inter2And2Helper bug [00:44] i was going to guess firewalls around the vps [00:44] oh right, missing username would be a different error [00:49] Good morning. [00:53] gotta say, ropemacs is da bomb! [00:54] when inlining a method that calls a function from another module, it adds the needed imports in the modules of the callsites [01:04] ah, i never managed to get it to work [01:05] Mh... I did need some non-trivial .emacs hacking here. [01:06] Here's what I did: http://pastebin.com/eiEVBJDJ [01:07] the python-ropemacs-hook is personal preference [01:09] I'm not sure why did this thing for on-startup loading, maybe on-demand loading was broken for me... [01:17] doesn't seem too bad [01:17] will have to try again at some point [01:19] poolie: huh, that Inter1And2Helper bug looks shallow, but also a bit depressing: [01:19] # XXX: not covered by tests, should have a flag to always run [01:19] # this. -- mbp 20100129 [01:20] s/source_repo/source/ looks like the obvious fix. [01:21] Also, in a probably unrelated issue, _get_rich_root_heads_graph looks suspiciously useless. [01:52] hm, thanks spiv === Ursinha-brb is now known as Ursinha-afk [03:41] so how should a --no-confirmations flag get down to the ui [03:41] poolie: via the context? ;) [03:42] hm, maybe :) [03:42] I've put up a patch for https://bugs.edge.launchpad.net/bzr/+bug/636930 btw, including test coverage. [03:42] Launchpad bug 636930 in Bazaar "Upgrading a repository fails with 'Inter1and2Helper' object has no attribute 'source_repo' (affected: 4, heat: 34)" [Critical,In progress] [03:42] should this be remembered in the uifactory, or in the context, or should it perhaps be somewhere else [03:43] extreme factoring would suggest there's a separate "user interaction policy" thing that is separate from the implementation of how to interact with the user [03:43] (legend) [03:45] and then that policy object would know whether to ask or not [03:46] Yes, I think that sounds about right. [03:46] I'd want to keep the policy object as simple as possible until more things use it. [03:47] But otherwise I think that sounds like a nice approach. [03:53] Hmm, maybe I should to fix http://bugs.python.org/issue9729 so that the test suite can pass on maverick. [04:06] could be good [04:06] there's another bug very slightly similar to that [04:06] just in bzr [04:06] which is something to do with socket.error being a bit odd [04:06] so not caught properly [04:06] perhaps because it's not a subclass of IOError or OSError [05:58] spiv, teddy bear? [05:58] i'm trying to decide if the user interaction policy layer should be a decorator layer on the uifactory [05:59] in some ways that's an easy way to intervene around all actions [05:59] * spiv listens [06:00] * spiv sits with arms sticking outwards a little bit, in his best imitation of a teddy bear ;) [06:00] :) [06:00] but it feels a bit wrong [06:00] Yeah, my initial reaction to that is it feels a bit weird. [06:00] if we do it, we could add a GenericDecorator that uses getattr to feed through to the underlying one [06:00] for instance you might like to look at the current state, or reset the state [06:01] So the decorator could intercept e.g. confirm_action and decide to ignore the decorated object and just return True if that was the policy? [06:03] So this is to save the individual uifactories from having to all implement the same "first, check the policy" logic? [06:04] right [06:04] how do you jump back a revision on a checkout? [06:04] and to make it easier to vary the policy [06:04] i can't believe i dont know how to do this [06:04] rryan 'bzr update -r -2' [06:04] for instance to have --confirmation=yes or --force options [06:04] https://bugs.edge.launchpad.net/bzr/+bug/397315 [06:04] Launchpad bug 397315 in Bazaar "break-lock should have a --yes, --force or --unconditional option (affected: 1, heat: 6)" [Medium,In progress] [06:05] facepalm, thanks! [06:05] So here's another use case that may or may not help to think about: [06:06] add the ability to selectively set a policy for just a specific question, e.g. "break_lock=yes/no/confirm" [06:07] I suppose the decorator would have to inspect the args passed to confirm_action (or whatever) to see if it matches the specific question. [06:09] (Hmm, I suppose a different variation is per-location or per-branch UI policy... I guess that's equally messy with the decorator or with every factory explicitly delegating to policy) [06:23] right, so we're going to want it somehow hooked up to the context, from which it can get the config object [06:23] i guess assigning a new object to ui.ui_factory feels like a very heavyweight action [06:23] to set the policy here [06:25] Yeah. [06:26] The decorator approach does sound like less code, which is a big plus. [06:30] What's the best way to maintain a changelog file with bzr? [06:51] lucidfox: well, there are two ways [06:51] you can do bzr log --format=gnuchangelog [06:51] or, you can manually commit to it and record those changes [06:58] Well, problem is [06:59] because of my "commit early, commit often" work style, automated log generation commits spams the log with one-line entries like "fix typo" or "add comments" [06:59] and the ChangeLog file is supposed to list *functional" changes, right? [07:00] right, so in that case i'd suggest probably maintaining it by hand, rather than automatically [07:30] Ok, patch submitted to http://bugs.python.org/issue9729. If upstream likes it I'll propose it for backporting to Python 2.6 in maverick, which will make our test suite a tiny bit happier. [07:32] It's a bit disconcerting fixing bugs in the Python standard library... the test suite doesn't seem anywhere near as comprehensive as I'm used to with bzrlib. [07:47] heh [07:47] feedreader was fun [07:47] "When the light of life has gone, no change for the meter. Then the king of spivs will come, selling blood by the litre." [07:47] they have a ton of tests but they're all sample data matched against the parsed form [07:48] indeed [07:49] so what do you think for a name? --yes? --jfdi? --no-confirmation? [07:49] --make-it-so [07:49] can't go wrong with Picard references [07:49] for bug 397315 [07:49] Launchpad bug 397315 in Bazaar "break-lock should have a --yes, --force or --unconditional option (affected: 1, heat: 6)" [Medium,In progress] https://launchpad.net/bugs/397315 [07:50] --thats-an-order [07:51] --crowbar [07:51] (lock breaking and all) [07:58] --yes sounds ok to me, it reminds me of fsck -y [08:05] hi all [08:05] :( i like --crowbar [08:07] poolie: why not a config var to define the ui itself ? [08:10] vila, how do you mean? [08:10] vila, istr that null output from script commands matches anything? [08:10] is that true? [08:10] it's just been confusing me [08:11] poolie: a yes_ui [08:12] null output matches anything... hmm, doesn't ring a bell, that would be quite a bug [08:12] mm it seems to be true though [08:13] vila, i agree, i think -O'ui=YesUI' should work [08:13] oh, wait, if *no* output is specified, then, yes, I think it's defined as meaning: "I don't care about the output", but that's not the same as specifying an output and getting none [08:13] that needs just a little more buildup and it's a somewhat longwinded way to get it [08:13] yes, that's consistent with what i'm seeing, but i think it's a bug [08:13] i recall discussing it [08:14] i think if you don't care, you should say '...' [08:14] poolie: but then you have to do it always, yeah may be we discussed that already [08:15] i should say i'm generally really super happy with scripts [08:15] I don't know if we have already enough data for that (nor if can define >/dev/null) [08:15] that's why i'm hitting bugs [08:15] :) [08:15] hehe [08:15] i think robert has a point they could create a pressure towards excessive blackbox testing [08:16] I fully agreed and still agree with that. I consider them a first rough shot to reproduce a bug [08:16] from there you drill down and write more precise tests [08:16] i think if you're going to write a blackbox test, you might as well write it this way [08:16] it's clearer and probably no slower [08:17] and when the bugs are shaken out, it should be no less precise [08:17] there's still the question of how we could test the command ui other than by full-stack tests [08:17] (how or whether) [08:17] https://bugs.edge.launchpad.net/bzr/+bug/637830 [08:17] Launchpad bug 637830 in Bazaar "null output from test script matches anything (affected: 1, heat: 6)" [Medium,Confirmed] [08:17] oh, for blackbox tests yeah, I was thinking more generally [08:17] no point making it _hard_ to write bb tests as a way to make people write unit tests [08:18] lol, no, that's not my point :) [08:22] vila, i think https://code.edge.launchpad.net/~mbp/bzr/ui-factory/+merge/34956 is now all ready to land [08:25] poolie: didn't I approved it ??? ... looks like it... weird [08:26] poolie: done [08:35] hmm, that's cleanup day :) I like it, go submitters, go ! [08:38] poolie: Why bzrlib.ui.testsupport instead of bzrlib.tests.ui (or something, but under tests) ? [08:45] heya GaryvdM [08:45] Hi bialix. [08:46] Gary, I've pushed small fix to lp:bzr-windows-installers branch, please, next time you'll build the installer use it [08:46] GaryvdM: btw, how's your own build host going? [08:47] bialix: I installed a 32 host, so I was not able to build the tbzr 64bit shell ext. So I'm gonig to have to start from scratch [08:47] vila, i actually did originally put it somewhere like that but i think python was getting confused by multiple things called ui [08:47] or at leoast i was :) [08:47] cleanup day? [08:48] GaryvdM: oops :-( [08:48] hi bialix, garyvdm [08:48] heya poolie [08:48] Hi poolie [08:48] ueah, I thought about that :) So how about bzrlib.tests.ui_utils ? [08:48] Hi vila [08:48] sure [08:48] hey GaryvdM, bialix ! [08:48] GaryvdM: which reminds me: why we still provides 32-bit bython version for 64-bit Windows? [08:49] vila: bonjour \o_ [08:49] i wouldn't mind having a rule that code outside .tests shouldn't call things inside it [08:49] aside of course from cmd_selftest etc [08:49] and this would advance that [08:49] poolie: yup, hence my remark about branchbuilder on the review comment [08:49] I'll probably build the 2.2.1 installer on the ec2 instance, and then have another go after that is done. I'll have to ask jam to revert the pyqt version. [08:50] GaryvdM: how hard it would be to have both 32 and 64 bits hosts? [08:50] bialix: Re 64bit python, - [08:50] bialix: at least twice as hard ? :-P [08:50] vila: exactly! [08:50] bialix: Re 64bit python, - I don't really know, I have not really looked at it [08:51] I have plans to get new computer soon, it should be 64-bit [08:51] bialix: My knowledge of the installer build is still small. I'm just following the docs.... [08:52] the only one question which bother me is: where to get the compiler for python extensions [08:52] bialix: good timing, 10.10 is nearly out :-P [08:52] vila: yep, I'd like to have dual boot machne [08:52] GaryvdM: which compiler are you using? [08:52] bialix: I was using VS C++ express, which is free beer, but not free speech. [08:53] bialix: even better: install 10.10 and use vms... way to go... ;) You will need a bit more RAM, but then you'll have both 32bits and 64bits windows available... [08:53] vila: I already have 32-bit Windows XP and pretty happy with it [08:53] bialix: you mean you'll get an *additional* computer ? [08:54] yes, new one [08:54] nice, have a look at synergy then, 2 computers is a nice setup, 2 mice, 2 keyboards... less nice [08:55] * vila looks around and see 5 mice 3 keyboards [08:55] hehehe [08:55] who's talking [08:56] I use synergy. [08:56] well, I use 1 kbd 1 mouse 99% of the time though [08:56] but there are cases were one PC needs a keyboard (boot, nasty problems) and 3 mices should just go in the attic :) [08:57] mice [09:01] vila i just got a magic touchpad today [09:02] it's pretty sexy [09:05] what's a magic touchpad ? [09:05] Oh, the apple one ? [09:05] yes [09:05] Can it call spells of 80 level? [09:06] :) [09:06] it can scroll both downwards _and upwards_ which my worn out old mouse couldn't do :) [09:07] poolie: yeah, I have a magic mouse (2 in fact :-p) it's a pity I had trouble configuring it properly for osx/vbox/lucid/synergys though :-/ [09:07] poolie: it works under maverick right / [09:07] spiv do you want to send your approved branches? [09:07] actually it's only on os x for now [09:08] poolie: ha. [09:08] i might try it on maverick later but my desktop pc has no bt adapter [09:08] oh nm, you did [09:17] vila i moved the uifactory stuff to tests.testui and it seems happy there [09:18] hehe, I thought about that too but didn't want you to say: oh, testui and test_ui are too close ;-p Please land ! [09:19] urk [09:19] they kind of are, but i think it's ok [09:21] sorry if it is a faq, probably I'm searching wrong keywords and cannot find the answer, how do I checkout a specific revision with bzr? [09:22] fargiolas: There are many ways to skin that cat [09:22] fargiolas: If you don't allready have a branch: bzr branch -r x [09:23] If you allready have a branch: bzr pull . -r x --overwrite [09:23] or [09:23] bzr revert -r x [09:23] or bzr update -r x [09:23] don't know if I have a branch, I ran brz branch lp:project-name [09:24] I still have to better understand bzr lexicon [09:24] I'm not sure what you want to do exactly. If you can tell me more, I can advise you better [09:24] you do [09:24] fargiolas: you want a copy of the tree as at a particular revision? [09:24] GaryvdM: I want to look at a project history looking at each commit changes [09:25] fargiolas: You do have a branch. do cd project-name; ls [09:25] if you're familiar with git, I'd like something like git log -p [09:25] 'bzr log -p' :-) [09:25] really, cool :D [09:26] as I said I'm pretty new to bzr, sorry :-) [09:26] fargiolas: no problem. [09:27] fargiolas: if you prefer fancy GUI, try 'bzr qlog' you need the qbzr plugin for that but it's worth it [09:27] vila: I'll try it, thanks [09:35] vila did you see spiv's neat "testdoc"? in lp:testdoc [09:36] prints a summary of the contents of a test module [09:36] i like '-f shiny' [09:36] poolie: yup, I need to use a dark backgrounf though or most of it is unreadble :) [09:37] ga <- fix typos [09:45] poolie: yeah, I'll submit them. I wanted to make sure the Inter1and2Helper fix got in because 2.2.1 is due quite soon. [09:46] How to launch bzr-gtk or bzr-explorer on Windows systems? [09:48] RomanMoz: If you installed with the windows installer, you should see a short cut in the start menu, and optionally on the desktop [09:48] RomanMoz: How did you install? [09:49] RomanMoz: From the console, you can run "bzr explorer" [10:05] hi [10:06] does someone here used bzr-git plugin? [10:11] hrw: sometimes [10:12] so far all my attempts ends with backtrace or other erros [10:13] bzr: ERROR: Invalid http response for https://github.com/hrw/linaro-armel-cross-toolchain.git/info/refs: Bad status line received [10:14] ok, found bug for it [10:14] bug 581933 [10:14] Launchpad bug 581933 in Launchpad Bazaar Integration "Imports from http://github.com fail, while ones from git://github.com work (affected: 1, heat: 6)" [High,Triaged] https://launchpad.net/bugs/581933 [10:15] hrw: Hi [10:15] hrw: "smart" http repositories aren't supported yet, but git:// should work for github repositories [10:16] 11:15 hrw@home:bzr-import$ bzr git-import git://github.com/hrw/linaro-armel-cross-toolchain.git [10:16] bzr: ERROR: exceptions.AttributeError: 'InterRemoteGitNonGitRepository' object has no attribute 'fetch_refs' [10:18] I prefer to create bzr branch with full git history but doing it by "git format-patch" + playing with bzr version of "git am" is what I do not want to play with [10:19] hrw: What version of bzr-git is that? Trunk should work. [10:20] *** 0.5.2-1 0 [10:20] 650 http://pl.archive.ubuntu.com/ubuntu/ maverick/universe amd64 Packages [10:20] hrw: Alternatively, you can have Launchpad do the import for you. [10:21] jelmer: where it is in LP? [10:22] hrw: The "Import a branch" link on the Code page of a project. [10:22] hrw: Or if you mean bzr-git itself, lp:bzr-git [10:23] ok, so time to create projects first [10:24] vila wow we're really going to need to see about giving better errors on failure [10:24] the assertionerror is not always super helpful [10:25] 'morning poolie, vila [10:25] hi there jelmer [10:26] poolie: ECONTEXT [10:27] jelmer: hey ! [10:29] morning all. [10:30] mgz: _o/ [10:30] vila there's not much explanation of which bit of the script it's unhappy about [10:30] but i just improved it slightly [10:30] I see pqm doesn't have news merge. [10:31] poolie: ha, will lineno help ? [10:31] it looks like only a few tweaks are needed to make '' not be a wildcard [10:31] mm, or just giving the command line [10:31] mgz: hmm yeah, I should probably file an RT to get it turned on. It only helps a bit, but that's still worth something. [10:31] that's not quite unambiguous [10:31] spiv you should [10:32] ok :) [10:32] spiv fixed bug 625574 during the night? [10:32] Launchpad bug 625574 in Bazaar "testtools details should not be shared between parameterised tests (affected: 1, heat: 9)" [Medium,Confirmed] https://launchpad.net/bugs/625574 [10:33] mgz: ha, thanks, I mentioned you to spiv but couldn't find back the mp... :) [10:35] spiv, mgz : I let you discuss it :-) [10:35] actually spiv i thought there was one already? [10:38] is this a known bug i sometimes get EBADF inside paramiko running parallel=fork? [10:38] adding a copy method was what Robert mentioned as well, just hadn't got as far as digging into what would need fixing where myself [10:38] probably thread leaks of some kind [10:40] thx [10:48] poolie: I dunno it a bug ws filed for it, but yes, it's more apparent now that the leaks have been fixed but it's an old one [10:49] poolie: the root cause is that when a client and server communicate via socket the events related to the socket being EOLed do not always occur in the same order [10:50] poolie: we catch EBADF on both the client side and the server side, but paramiko use its own private thread, where EBADF, sometimes, occurs [10:50] poolie: IIRC it doesn't make the test fail thoug, it's just annoying to have it in the output at all :-/ [10:53] ah ok [10:53] vila ah i see the naughty test_conflicts blackbox tests, but not in that directory :) [10:53] [10:54] hm [10:54] so they are nice and readable [10:54] they are kind of using the command line to do setup, which will tend to be slow [10:54] poolie: yeah, I should rewrite them into more precise tests as hinted earlier :) [10:54] and i think was lifeless's objection to adding this facility [10:55] just thinking out loud, not trying to give you a hard time [10:55] no no, np [10:55] I was in exploratory mode when I wrote them so they fitted the bill: how to create such and such conflicts [10:56] some other things it's teaching me [10:56] AFAIK there are the most complex setups we have and spiv and I weren't fully happy about the parametrization either [10:56] we're a bit inconsistent about stdout vs stderr (not totally news) [10:57] and also, a universally respected -q would be good for these tests [10:57] also apparently -1 doesn't work with --parallel [10:58] no, at best it could stop one thread (well process) [10:59] these tests are a good example of why I didn't care about command output... until the very end, so adding '...' for every line will make them *far* less readable [10:59] actually -q is pretty good [10:59] on the other hand, they are not here to stay... [11:00] adding it to every bzr invocation in the script ? wfm [11:00] to the ones you don't care about [11:00] an alternative would be to add an option to the run_script method to keep the actual behavior [11:01] hi guys, i'm trying to write a plugin to add credentials into gnome keyring in the format that bzr-gtk expects [11:02] but i'm getting lost in the code somewhat - can anyone point me to how to get the credentials stuff from a URL? [11:03] Glenjamin: look into bzrlib/transport/__init__.py split_url and unsplit_url [11:03] Glenjamin: be aware that the credential stores doesn't provide any facility for storing new credentials [11:04] Glenjamin: the rationale is that keyring, ssh-agents are best fitted for that [11:04] the gnome-keyring bit in bzr-gtk uses a hash of attributes to store [11:04] the gnome seahorse program doesn't let you set them [11:06] so i can't see a way to non-programatically set the password [11:07] I see, but still :) [11:07] * vila searches [11:08] so the good news is that -q does consistently suppress most stuff [11:09] i suspect there are still some where it's not correct [11:11] Glenjamin: meh, seahorse let you create passwords or keyrings or ssh keys or pgp keys, what are you trying to create ? [11:12] http://bazaar.launchpad.net/%7Ebzr-gtk/bzr-gtk/trunk/annotate/head%3A/keyring.py#L60 [11:13] you can create a named password, but you cant set the attrs dictionary from seahorse - although it will display it for existing passwords [11:14] Glenjamin: did you look at oython-gnomekeyring ? [11:15] yes, it has bindings to let me store a password [11:15] i managed it manually [11:15] but now i want to be able to do "bzr gpass some://url" and have it prompt me for username/password, and store it [11:16] i'm struggling to convert the URL argument into the relevant bits however [11:16] Glenjamin: haaa, ok, so you don't try to store them in authentication.conf ? [11:17] as ideally i'd like to be able to plug in the same url that any other command would take - in this case it's a custom directory service which maps to bzr+https [11:17] vila: nope, if the password is in the keyring it seems to pick it up [11:17] Glenjamin: then look at bzrlib/transport/__init_.py [11:17] that would be a nice feature [11:18] i'm intrigued as to how jelmer manages to get the passwords into keying - as he's the author of this bit of bzr-gtk [11:18] Glenjamin: that's the trick, it doesn't, he just uses the ones that are there [11:18] s/it/he/ sorry jelmer :-/ [11:18] but they'd have to get there somehow, no? [11:19] Glenjamin: other apps can add them [11:19] yay, done, and not too tedious to change [11:19] vila https://code.edge.launchpad.net/~mbp/bzr/scripts/+merge/35386 - no rush [11:19] Glenjamin: circa line 1354 there is one example [11:19] thanks for the reviews! [11:19] poolie: ok [11:19] lunch time here, bbl [11:28] annoyingly, parsing the url just gives me bzr+https. which isn't much use. Since this is only for one repo i'm just going to make a more specific command. :( [11:51] i've run into a weird error. I've got an https smart server running, and http redirecting to https. if i try and do a command to bzr+https, i get an error about 401 being an invalid response. If i do bzr+http it redirects to https and prompts me for my password. :s [11:56] and it only happens with pycurl [13:09] hmm. how do i check if a revision is valid from a revision_dwim object? [13:13] Glenjamin: what do you mean by 'only bzr+https' ? Care to paste some code demonstrating that ? [13:13] !paste [13:13] For posting multi-line texts into the channel, please use http://paste.ubuntu.com | To post !screenshots use http://tinyurl.com/imagebin | !pastebinit to paste directly from command line | Make sure you give us the URL for your paste - see also the channel topic. [13:19] knittl: hmm, I don't think you can, the dwin processing is about trying various kind of revisions until one succeeds, see bzrlib.revisionspec circa line 325 [13:19] vila: well, i want to test if it's 0 [13:19] so -r0 [13:20] and print revision outputs [13:20] knittl: what's wrong with -rnull: ? [13:20] vila: it crashes [13:20] * [13:21] knittl: what crashes ? [13:21] bzr [13:21] ok, doing what ? [13:21] inventory -r0 [13:21] inventory -r0 . works (giving filelist) [13:22] but it should output nothing right ? [13:22] yes [13:22] file a bug, probably pretty shallow [13:22] or a warning 'r0 does not have an inventory' or something like that [13:22] vila: i try to fix it myself [13:23] knittl: good, thanks ! [13:23] that's why i asked how to check the revision ;) [13:23] knittl: but once the bug is filed you can assign to yourself and everybody will know it's a know one if they encounter it [13:24] nobody is as stupid as i am and will try to show the inventory of r0 ^^ [13:24] yada yada [13:25] I'm surprised it didn't show up in other contexts... and that we don't have a test for that [13:25] if type(dir_ie) is type(None): return; [13:26] in inventory.py entries(self).descend(dir_ie, dir_path) [13:26] seems to fix it [13:27] in descend ? yeah, I wsa there too (note that we use the 'if dir_ie is None' idiom instead [13:27] that did not work [13:27] but it was 4am when i tried last [13:27] :D [13:27] oh ok, works too :) [13:29] should this go inside descend or should i check self.root before calling descend? [13:29] checking root.self will be called only once instead of each subdirectory [13:30] knittl: I think it's an old code path, http://paste.ubuntu.com/493618/ works [13:30] lol, ok [13:31] knittl: well, you did the hard work :) Please send a merge proposal ! [13:32] although entries(self) says "this may be faster than iter_entries" [13:32] no, it's bogus, it breaks the test suite [13:33] ok, iter_entries works too [13:33] knittl: no it's not, it breaks the test suite :) [13:34] what? entries()? [13:34] hu [13:34]