[09:01:42] 10Quarry: Add a list/table of popular queries - https://phabricator.wikimedia.org/T71266#1444241 (10Edgars2007) @He7d3r don't think this is a good idea to get top N queries by number of times they were executed. When one is testing the query, he probably executes it //to much//. But I support the idea itself to... [11:26:30] 10Quarry: Add a list/table of popular queries - https://phabricator.wikimedia.org/T71266#1444564 (10He7d3r) ...maybe the most accessed (i.e. most page views)? [13:46:50] morning halfak :) [13:50:08] o/ Ironholds [13:50:17] how goes? [13:51:16] Lazy day today. Just rolled out of bed. [13:52:11] I went to a b-day party last night for one of Jenny's labmates and drank most of a bottle of sake in order to preserve Jenny's DD status. [13:52:56] halfak, I don't think that's how it works, but I'm glad you had fun :D [13:53:08] unless you mean Jenny has some irresistible draw towards Sake, or something [13:53:29] "noo, let me save you!" *jumps, takes a bulleW^shot of sake in the chest* [13:54:20] :D It's a party! Any other maneuver would have been inappropriate. [13:54:38] But seriously, I can't let things go to waste! [13:54:54] * halfak remembers his grandma talking about starving children [13:55:22] your grandmother fed children sake?! [13:55:45] I don't know that she didn't! [13:58:46] halfak: I'm technically off today - we should figure out plans for the research tools workshop :) [13:59:13] Not today tho [13:59:18] I'll think about it and stuff [14:00:07] YuviPanda, hokay. I think we're going to have substantially fewer researchers present than we did at the last Wikimania [14:00:22] halfak, also, want to hear something horrifying? [14:00:33] So I gave my talk in Denmark on unit testing and software engineering in R [14:00:39] and I got the reply from a member of the R Core team [14:00:44] "we don't need unit tests, we have examples" [14:00:44] But I'd like to pull in people like He7d3r and danilo because they do analytics work for other Wikipedians all the time. [14:00:52] * Ironholds twitch [14:01:27] Ironholds, if you can execute the examples, then they might be useful for supplementing unit tests. [14:01:46] I have an extension in my testing framework for running examples that appear in the docs and making sure the outputs are right. [14:01:59] But my tests are far more numerous than my examples. [14:02:00] halfak: yeah I titled it 'how to do research as a bolunteer' [14:02:05] Or something like that [14:02:10] halfak, yeah, I'm writing a thing about the problem with examples [14:02:12] you can execute them! [14:02:21] except they, by definition, can never throw errors [14:02:24] so you can only test positives [14:02:33] YuviPanda, I wonder how we could convince this crowd that this applies to them. [14:02:39] I can test rnorm(12) generates a number, I cannot test rnorm(12, sd = -3) consistently bugs out [14:02:45] YuviPanda, Maybe we want a post to wikitech-l [14:02:57] halfak: and the research list maybe [14:03:05] YuviPanda, +1 [14:03:21] Ironholds, indeed. [14:03:26] * YuviPanda thinks about conversations at the beginning of quarry about democratizing research [14:03:31] Ironholds, Unless you capture the error in the example. [14:03:39] YuviPanda, +1 [14:04:06] I want to be able to convince a large crowd of people that they are both interested in research *and* maybe already doing it. [14:04:13] The word "research" is weird. [14:04:20] Yeah [14:04:29] And I expect a lot of people don't associate with it. [14:04:37] 'Having questions and finding answers to them' is a bit unweildy [14:04:41] Maybe not. Either way, I think we want a wide net. [14:04:51] Yes [14:05:11] Ironholds, thinking about having errors in examples sounds silly. [14:05:32] Also -- this is just a thing -- but why is it that examples are the last thing in a man page? [14:05:44] idle fancy from the R devs, I guess? ;p [14:05:47] I almost always want an example call when I man [14:05:51] but yeah, errors in examples are dumb, and so /we don't do that/ [14:05:58] +1 [14:06:02] and that means that there are no unit tests in a looot of packages for error handling [14:06:09] not that they use unit tests [14:06:11] Of course [14:06:19] something like 28% of R packages are using any kind of recognizable test [14:06:39] and to make matters worse, a lot of examples are wrapped in \dontrun{} tags because CRAN fails your package if an example takes >N seconds, to preserve hardware time [14:06:46] * Ironholds throws hands up [14:06:51] so many things for me to fix [14:07:42] Heh [14:08:17] oh well. Starting to fix some of them (actually stole an idea from Perl. Mwahahah) [14:08:29] Perl is full of good ideas [14:09:02] Perl is a reflection of the universe. It has all the things and good ideas are types of things. [14:11:26] yup! [14:11:30] so, they have meta-cpan [14:11:45] there's a module you can install packages through and optionally upload the install results and your system info to a centralised table [14:12:08] so the R equivalent would be a little table you can click through to see precisely what happens if, say, you install ggplot2 1.9.0 on R 2.9.3 on a 64-bit Windows machine [14:12:37] CRAN just checks latest version, which is GREAT except for the bit where people are still running 2.9 [14:12:58] wb, halfak :) [14:13:34] Woops. [14:13:38] :S [14:13:40] crtl-Q is really close to ctrl-W [14:14:58] ahah [14:15:29] * halfak spools up Hail Mary Mallon. [14:15:50] Ironholds, I doubt such a thing exists for python. [14:15:53] But I want it. [14:16:09] It would be really nice to know what OS/version pairs are failing to install my packages. [14:16:58] halfak, to my knowledge it doesn't but it would be pretty doable [14:17:08] I'm actually probably writing the API and web service in Python [14:20:22] Makes sense. Either that or Ruby, but python is better for sciency things assuming that you might do some sciency things in this service. [14:21:31] well, I'm simply more familiar with Python [14:21:51] Sounds like a fine reason to me. [14:21:52] Ruby is basically magic to me (not voodoo. C is voodoo. Because voodoo is magic with unexpected death) [14:54:57] Hello, scientists. [14:58:50] gah [14:59:01] I go away for ~two weeks~ on holiday and come back finding my UDFs are still not live [15:20:05] o/ guillom [15:20:48] * halfak adds 'scientist' to his ping word list. [15:30:13] Hello, Senior Research Scientist Aaron Halfaker. [15:33:53] \o/ :) [21:05:29] Ironholds, get into research group! [21:05:36] halfak, see etherpad! [21:05:44] I can't make it - gotta get head down on a c-level task :/ [21:05:59] Hokay. Have fun :( [21:06:19] I will have as much fun as I have daily in my job [21:06:39] fun, fun, fun, til daddy takes the T-bird away? [21:08:08] the T-bird was already taken away about a year ago, unfortunately [21:08:17] Ironholds: I'm sorry :( [21:42:50] heh; I missed the research meeting, didn't I? [21:42:58] * guillom is not used to having meeting on Fridays. [21:43:02] guillom, still going [21:43:04] get in here [21:43:08] ok! [22:36:21] I'm off! The next time I connect will be from Mexico [22:36:22] o/ [22:36:53] Safe travels!