Perl 6 - the future is here, just unevenly distributed

IRC log for #opentreeoflife, 2014-08-11

| Channels | #opentreeoflife index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
02:30 towodo joined #opentreeoflife
02:44 towodo joined #opentreeoflife
12:16 kcranstn joined #opentreeoflife
12:30 jimallman joined #opentreeoflife
12:34 towodo joined #opentreeoflife
13:16 kcranstn joined #opentreeoflife
14:36 kcranstn thes viz links on the developer resources page don’t seem to be working : http://files.opentreeoflife.org/other/synth_viz_dyn.html?ottolID=304358
14:41 kcranstn Cross-Origin Request Blocked: The Same Origin Policy disallows reading the remote resource at http://ot9.opentreeoflife.org/treemachine/ext/GoLS/graphdb/getNodeIDForottId. This can be fixed by moving the resource to the same domain or enabling CORS.
14:41 scrollback joined #opentreeoflife
14:43 kcranstn (was trying to demo the viz for someone interested in using d3 for phylogenies)
14:58 jimallman kcranstn: i’ll look into this. not sure why we’re using ‘ot9’ URLs here...
14:59 kcranstn thanks!
15:22 jimallman kcranstn: fyi, these are working now. i switch the synth- and taxa-viz scripts to use more robust URLs on api.opentreeoflife.org
15:22 kcranstn awesome, thank you!
15:22 jimallman i’ve modified the live files in-place and made the corresponding changes in the opentree repo.
15:32 kcranstn joined #opentreeoflife
15:59 josephwb joined #opentreeoflife
16:47 towodo_ joined #opentreeoflife
16:55 josephwb1 joined #opentreeoflife
16:58 josephwb joined #opentreeoflife
19:55 towodo joined #opentreeoflife
20:16 towodo joined #opentreeoflife
20:37 josephwb you there @jimallman?
20:38 jimallman hi, i’m in a call but will be done shortly.
20:38 josephwb ok
20:38 josephwb this might also benefit from feedback from towodo as well
20:39 towodo this = ?
20:39 josephwb i was going to wait for Jim
20:39 towodo o
20:39 josephwb I've got the treemachine getMRCA plugin working and tested
20:40 josephwb it now has a additional parameter: taxonomyOnly (Boolean) i.e. should only taxonomy relationships be traversed
20:40 jimallman i’m here!
20:41 jimallman cool. have you deployed this to devapi? or should i?
20:41 josephwb my question: what would you like as the default value of "taxonomyOnly"? I vote for TRUE
20:41 josephwb but then I was wondering if this should be two separate services
20:41 jimallman sure, let’s try that. (this will give immediate results on devtree, even without any changes on my end)
20:41 josephwb "blastAgainstTaxonomy"
20:41 jimallman hmm
20:41 josephwb "blastAgainstSyntheticTree"
20:42 josephwb more clear
20:42 jimallman ok, that does make sense
20:42 josephwb cody likes to break up such services
20:42 towodo I thought blast was passe
20:42 josephwb i don't really care
20:42 josephwb towodo those names were not permanent
20:43 towodo right, ok
20:43 josephwb "getMRCATaxonomy"
20:43 jimallman the existing service is called ‘getDraftTreeMRCAForNodes’
20:43 josephwb "getMRCASynthesis"
20:43 josephwb yes
20:43 josephwb yes, jimallman
20:43 jimallman hm, or “getTaxonomyMRCA” (less awkward capitalization)
20:43 towodo have we given up the idea of using tree parameters inside of treemachine? i.e. multiple simultaneous tree, such as taxonomy, source trees, and one or more synthetic trees?
20:44 jimallman or “getMRCAViaTaxonomy”?
20:44 josephwb towodo: multiple synth trees is a goal, but no one is working on it at the moment
20:45 josephwb i don't think anyone is thinking about multiple taxonomies
20:45 jimallman right, this comes up periodically and would be great (if potentially confusing, if someone chases a URL to old synthesis)
20:45 josephwb yes!
20:46 jimallman josephwb: well, we were talking in today’s call about offering more information about past OT taxonomy versions. is it reasonable for someone to want to explore them?
20:46 jimallman (easily, i mean)
20:46 josephwb i don't think it is undoable
20:47 josephwb possibly a hackathon task?
20:47 jimallman !
20:47 josephwb maybe
20:47 jimallman simple exploration of taxonomy is a long-standing request… would be neat to hack our way to it.
20:48 josephwb i think under a single taxonomy, it might not be too painful, but multiple taxonomies may be a nightmare
20:49 jimallman (i’ve planned for taxonomy browsing in URL conventions and routing, we just need to teach the server to generate arguSON and the viewer to show OT taxa properly)
20:49 josephwb anywho, what do you jimallman and towodo think about the whole MRCA dealie? One service, or two?
20:49 jimallman it seems like a consumer would clearly want one or the other, right?
20:50 josephwb that is how i am starting to feel
20:50 jimallman so i’m leaning toward two methods
20:50 towodo I wouldn’t worry about multiple taxonomies
20:50 towodo but being able to view the current one would be nice
20:51 josephwb view what?
20:51 jimallman re “two methods”: ...otherwise i fear someone would ignore the optional ‘taxonomy’ flag and not realize they’re seeing a critical default behavior
20:51 jimallman view the current taxonomy, i think
20:51 josephwb agreed
20:51 towodo I wouldn’t worry about being about to view multiple taxonomy versions at the same time, but it would be nice if one could view the current taxonomy version
20:51 jimallman yes
20:52 jimallman can they get or reconstruct an old taxonomy version “the hard way”?
20:52 towodo re synth + taxo, I’m worried about orthogonality with many different methods.
20:52 josephwb "view the current taxonomy version" you mean a newick tree?
20:52 towodo no
20:52 towodo I mean like a taxonomy browser, like we used to have (argus)
20:52 jimallman i think we’re talking about using the tree explorer (eg tree.opentreeoflife.org), but for taxomony instead of synth-tree
20:52 josephwb oh, ok
20:53 jimallman as i said, i’ve planned for this in web2py
20:53 towodo I think we might get better orthogonality if many methods would take a ‘tree designator’, maybe a string that indicates taxonomy vs. synthetic tree
20:53 jimallman re: orthogonality, i’m not sure what your concern is… i should review the APIs and it’ll probably come to me.
20:54 towodo otherwise we could potentially hae 2N methods
20:54 jimallman ah, gotcha
20:54 towodo my concern is an explosion of methods
20:54 towodo mrca is one, view is another, subtended tree is another, I bet there are others
20:54 jimallman is this on the notion that treemachine would provide analogous services for the OT taxonomy (as a tree), instead of putting those methods in taxomachine?
20:55 jimallman i see what you mean, there’s a sort of assumed context (the current synthetic tree) for the existing treemachine methods.
20:55 towodo that is one option, and it could be more economical in terms of duplicated or reimplemented code
20:55 josephwb so, for the two MRCA services: how about 1) "getTaxonomyMRCA", and 2) "getSynthesisMRCA"?
20:56 towodo but I was just arguing against having two services. does my argument not hold water?
20:56 towodo getMRCA(treedesignator)
20:57 towodo Or, put getTaxonomyMRCA in taxomachine ? was that what you were thinking?
20:57 josephwb works for me. i've never designed such things, so I will just do what you guys think is right.
20:58 josephwb re taxomachine: yes, that is an option
20:58 josephwb the methods may exist in there (they should)
20:58 towodo but we currently have no way for taxomachine and treemachine to share code, other than ot-base…
20:59 josephwb a worry is if taxomachine and treemachine have different taxonomies
20:59 towodo like they do now
20:59 * jimallman is on the phone, back in a miniute
21:01 jimallman if treemachine “knows” the taxonomy (at least, the one used in the current synthesis), it seems easier to keep the analogous behavior in treemachine than to teach taxomachine lots of new tricks.
21:01 jimallman and it covers josephwb’s concern about the two becoming out-of-synch
21:02 josephwb i feel like this is swinging back and forth :-S
21:03 josephwb it is possible to report the taxonomy version on check
21:03 josephwb i am looking now at taxomachine to see if it is all done there already
21:04 * jimallman is arguing all sides, just to think them through (sorry)
21:04 josephwb taxomachine does have some getMRCA code
21:08 kcranstn joined #opentreeoflife
21:08 josephwb to be honest, getting the taxonomy MRCA from traversing a graph DB seems a little overkill.
21:08 josephwb but it works
21:08 jimallman :D
21:09 josephwb kcranstn: we are discussing the "getTaxonomyMRCA" and "getSynthesisMRCA"
21:09 josephwb i have it working, but we are discussing whether it should be 1 service or 2
21:09 josephwb OR
21:09 josephwb treemachine + taxomachine
21:10 kcranstn just a sec
21:10 kcranstn reconnecting
21:10 kcranstn joined #opentreeoflife
21:11 kcranstn just had my pupils dilated, and having to increase font size
21:12 kcranstn browsing history
21:13 josephwb what is your take, ilbot3?
21:14 kcranstn joined #opentreeoflife
21:15 * jimallman is going to create a yesbot, just to agree with everything he says. (“Great idea, jimallman!”)
21:15 josephwb jimallman: the current service is "getDraftTreeMRCA", which is pretty descriptive. Perhaps I should undo what I did, because "taxonomyOnly" is confusing for "getDraftTreeMRCA"
21:15 kcranstn it seems clearer for the user to have two methods, but I do see towodo's point about number of methods increasing
21:16 kcranstn I tend to lean towards clarity
21:16 kcranstn it is hard for people to grasp the types of content we have
21:16 jimallman good point
21:17 kcranstn we can always put version numbers in the synthetictree and taxonomy methods
21:17 josephwb "clarity" = 2 methods, or "clarity" = fewer methods?
21:17 jimallman josephwb: agreed about the confusing flag vs. method name, IF taxonomy=true means we just traverse taxonomy and ignore phylo relationships. i originally thought it meant we’re traversing the synthetic tree, but disregarding all but OT taxa for the MRCA.
21:17 josephwb oh, i guess you are voting for 2
21:17 kcranstn I think two methods are clearer
21:17 josephwb i kinda like towodo's suggestion: getMRCA(tree_source)
21:18 kcranstn I could be convinced
21:18 josephwb where "tree_source" must be supplied (no mistakes)
21:18 kcranstn yes
21:18 kcranstn I like that
21:18 jimallman agreed, this works, if we’re planning on moving toward this for all tree operations
21:18 josephwb by "no mistakes", I mean the user does not intend the default parameter values
21:18 jimallman (which is jar’s position, i think, and one i’m leaning toward)
21:19 towodo tree_source could just be the string ‘taxonomy’ or ‘synthetic’, but it’s then available as an extension point, e.g. for versioned trees
21:19 jimallman yes! we’ve already done some of this, where ‘current’ (or somesuch) means the latest synth or taxonomy
21:19 kcranstn an elegant solution
21:20 josephwb oh, good point
21:20 towodo we are going to have the same problem for the subtended subtree method, yes?
21:20 jimallman assuming we want to offer this, yes
21:20 towodo that is, it currently gets a subtree of the synthetic tree, but it makes just as much sense to get a subtree of the taxonomy?
21:20 jimallman (there’s a viral quality to this change, i think, with a real payoff once it’s complete. but it might take awhile.)
21:21 towodo (that’s ok, I like doing things incrementally)
21:21 josephwb if we are planning a bunch of taxonomy features, it seems like taxomachine should do it (sorry for swinging the other way)
21:21 jimallman agreed, and if we plan it right, then operations can apply to the synthetic tree, the taxonomy, or any source tree. which is nice.
21:22 towodo it’s not specifically a taxonomy feature, it’s a tree feature
21:22 jimallman i guess i think of taxomachine as the *builder* of the taxonomy, and treemachine as the navigator and explicator(?) of trees in general.
21:22 jimallman assuming that said tree is in its graph db
21:23 josephwb we (I) do everything without ever using taxomachine
21:23 towodo … except that taxomachine doesn’t build the taxonomy, smasher does… taxomachine is the name server (TNRS)
21:23 josephwb so, taxomachine is *only* a TNRS?
21:23 josephwb that's fine
21:23 towodo currently that’s the case, I believe.  it’s not necessarily true
21:24 towodo ah… just thought of something… the taxonomy in taxomachine has more taxa in it than the one in treemachine
21:24 josephwb good point
21:24 towodo it has the incertae sedis taxa
21:25 towodo so in the current setup we really do want to have all taxonomy related methods go through taxomachine
21:25 jimallman hm, yep (or retain all taxa in treemachine, flagged as not-for-synthesis)
21:25 towodo but we could still have the tree_source parameter. either as a sanity check or as an extension point
21:26 towodo retaining the taxa is too big a change to treemachine
21:27 josephwb * head is spinning *
21:27 towodo ideally, we allow taxonomy method calls to go through either in taxomachine or in treemachine, so we have flexibility
21:27 towodo and to deal with version skew issues
21:28 jimallman * head is spinning *
21:28 josephwb i will just code this up in treemachine for the moment so jimallman has something to work with
21:28 towodo I said ideally.
21:29 jimallman thanks! :)
21:29 jimallman i think this is a great conversation for the larger s/w group. it clearly has interesting ripples.
21:29 josephwb i will make up my own mind whether it is one service or two, because frankly you guys have only confused me
21:29 jimallman ok, but it should be two
21:29 josephwb +o(
21:35 josephwb jimallman: expect a service called "getTaxonomyMRCAForNodes" soon. We can argue over the names/structuring later.
22:05 josephwb ok jimallman, it is pushed
22:06 josephwb i've also noticed where/why getDraftTreeMRCAForNodes was dying. Will fix that when i get home
23:09 josephwb joined #opentreeoflife

| Channels | #opentreeoflife index | Today | | Search | Google Search | Plain-Text | summary