Perl 6 - the future is here, just unevenly distributed

IRC log for #opentreeoflife, 2014-08-28

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

All times shown according to UTC.

Time Nick Message
00:59 kcranstn joined #opentreeoflife
07:11 mtholder joined #opentreeoflife
07:48 travis-ci joined #opentreeoflife
07:48 travis-ci [travis-ci] OpenTreeOfLife/phylesystem-api#585 (issue101 - 3c05135 : Mark T. Holder): The build has errored.
07:48 travis-ci [travis-ci] Change view : https://github.com/OpenTreeOfLife/phylesystem-api/commit/3c0513559421
07:48 travis-ci [travis-ci] Build details : http://travis-ci.org/OpenTreeOfLife/phylesystem-api/builds/33773932
07:48 travis-ci left #opentreeoflife
07:49 travis-ci joined #opentreeoflife
07:49 travis-ci [travis-ci] OpenTreeOfLife/phylesystem-api#587 (master - 3c88e7c : Mark T. Holder): The build passed.
07:49 travis-ci [travis-ci] Change view : https://github.com/OpenTreeOfLife/phylesystem-api/compare/693837e15a43...3c88e7c3adab
07:49 travis-ci [travis-ci] Build details : http://travis-ci.org/OpenTreeOfLife/phylesystem-api/builds/33773949
07:49 travis-ci left #opentreeoflife
07:51 travis-ci joined #opentreeoflife
07:51 travis-ci [travis-ci] OpenTreeOfLife/phylesystem-api#588 (issue101 - da9de18 : Mark T. Holder): The build has errored.
07:51 travis-ci [travis-ci] Change view : https://github.com/OpenTreeOfLife/phylesystem-api/commit/da9de183b7b5
07:51 travis-ci [travis-ci] Build details : http://travis-ci.org/OpenTreeOfLife/phylesystem-api/builds/33774152
07:51 travis-ci left #opentreeoflife
08:56 mtholder joined #opentreeoflife
08:56 mtholder left #opentreeoflife
08:56 mtholder joined #opentreeoflife
10:58 mtholder joined #opentreeoflife
12:45 kcranstn joined #opentreeoflife
13:36 towodo joined #opentreeoflife
15:38 josephwb joined #opentreeoflife
17:27 josephwb working on updating treemachine plugins
17:27 josephwb based on cody's info: https://github.com/OpenTreeOfLife/opentree/wiki/Proposed-API-changes-2014-08
17:28 towodo Don’t you think the design should get reviewed first? Or do you not care if you have to change it a 2nd time?
17:31 josephwb that's why I am here.
17:31 josephwb how should we go about reviewing this?
17:31 josephwb (by the way, I am on a separate branch, so I will not break anythign)
17:32 josephwb e.g. Cody proposes this:
17:32 josephwb getTaxonomyVersion -> ott_version
17:32 jimallman we’ve scheduled a sub-group hangout for this discussion, Fri. 11am
17:32 josephwb i imagine we do not want "ott" in there
17:32 josephwb just "taxonomy"
17:33 towodo Well, we should form a plan as to what the requirements are, when the review and discussion period will be, how deployment will happen, what the sunset period will be for the old API, the expected longevity of the new one, and so on.
17:34 towodo We can’t design something in 2 days and expect to be able to live with it for 6 months.
17:41 josephwb is it kosher practice to have two services be named the same thing if they are coming from different sources? there is a proposed "taxon_info" for both treemachine and taxomachine.
17:42 josephwb *i don't know how these things are done*
17:44 towodo look at other web APIs to see how other people do things
17:44 kcranstn http://www.gbif.org/developer/summary
17:44 kcranstn http://eol.org/api/
17:44 towodo I’d say dup method name is ok as long as the arguments and return values are compatible, which means probably not
17:46 towodo found this via web search https://blog.apigee.com/
17:46 towodo (has a lot to say about versioning https://blog.apigee.com/taglist/versioning )
18:38 codiferous joined #opentreeoflife
18:41 josephwb hey codiferous
18:41 codiferous hey
18:43 josephwb one thing that is confusing me about the proposed changes involves draftTreeForNode and subtree
18:43 josephwb one service you give it a bunch of tips
18:43 josephwb the other, you specify a starting internal node
18:43 codiferous yeah, i proposed combining them since they both produce subtrees
18:44 josephwb right
18:44 codiferous depending on the arguments you provide, you get a different type of subtree
18:44 josephwb i get that
18:44 josephwb but should we combine them?
18:44 josephwb if you specify 5 tips to subtree, do you get a tree back with 5 tips?
18:45 josephwb or, the 5 tips you specified + all the intervening taxa?
18:45 codiferous if you specify 5 tips in the "tips_for_pruned_subtree" argument (or whatever, that's a made up name), then you get a pruned subtree
18:45 kcranstn pruned = yes / no
18:45 kcranstn ?
18:45 josephwb ya lost me
18:46 josephwb 5 tips, or 5 tips + intervening taxa?
18:46 kcranstn should there be an option for pruned (yes/no)
18:46 josephwb e.g. duck, human, turtle, sunflower
18:46 codiferous sure
18:46 codiferous you could do that any number of ways
18:46 josephwb these seem to be quite different things
18:46 josephwb i may want all birds (specify internal node)
18:47 codiferous i would imagine using the getMRCA function to get a node id, then asking for the complete subtree for that node
18:47 codiferous two different services
18:47 josephwb or, i might have data for 5 birds, and want a pruned tree of just those taxa
18:47 codiferous i imagined the subtree services just implements queries that produce subtrees, we could have several different kinds
18:48 codiferous one case is pruned, in which case you have to provide tips. another case is complete, in which case you provide a root node. those are the cases we currently support
18:48 kcranstn I suspect there will be lots of input from the hackathon on how best to implement this
18:48 josephwb i understand now, thanks
18:48 codiferous i expect we will have new methods requested at the hackathon
18:48 josephwb but quite different things
18:48 codiferous yes
18:49 josephwb a little confusing in names
18:49 codiferous they could easily be implemented in different services (the way we do it currently)
18:49 josephwb a "subtree" that starts from a starting node is one thing
18:49 josephwb that is easty to combine with "draft_tree"
18:49 towodo just entered some thoughts about an agenda for tomorrow in the ‘weekly software calls’ google doc
18:49 josephwb the arg is the start node; root is not provided
18:50 codiferous yep
18:50 codiferous that would work
18:50 josephwb in fact, "draft_tree" already has this
18:51 josephwb "getSyntheticTree", as it stands currently
18:51 codiferous yep
18:51 josephwb i guess it is good to differentiate "pruned" vs. "subsetted" trees
18:52 josephwb subset = complete below a certain node
18:52 kcranstn thanks, towodo
18:52 josephwb pruned = subset of taxa below a node
18:53 codiferous hm? how is subsetted different from pruned?
18:54 josephwb subset = start at Aves, get all descendant taxa
18:54 josephwb pruned = give me a tree of duck, hummingbird, ostrich
18:54 josephwb say mrca, differemnt sized trees return
18:54 codiferous ah ok
18:54 josephwb former is "complete" below Aves, latter is not
18:55 josephwb maybe not the best names if it is not clear!
18:58 codiferous regarding our apis in general, in the long term, i think the built-in neo4j webservice extensions are not going to serve us well
18:58 codiferous in the short term, they're what we have to work with
18:58 josephwb yeah...
18:59 codiferous we can't do paging, we can't control much about http or serialization, and we don't have control of url structure (although apache redirects allow us to get around that)
19:00 codiferous so, in terms of designing a really robust, well-organized, and rich api, i think it's a little premature
19:00 towodo I’d be happy to transfer what I know about ‘unmanaged’ plugins
19:01 codiferous that would be helpful. that seems like the most promising direction to move in, but it's going to take a while
19:01 codiferous my current interests are in cleaning up the existing api so it is less confusing and more consistent before the hackathon
19:03 codiferous i expect we will get a lot of feedback during the hackathon, i think we will get more valuable feedback if we clean up things that we already know are pain points
19:04 josephwb trying to switch before the hackathon does not seem feasible
19:06 codiferous some things could be done. renaming methods and classes, and removing unused services, would take less than an hour
19:06 codiferous other things may not be feasible
19:08 josephwb yes, i am eager to do that
19:10 josephwb at least for treemachine
19:12 codiferous i'm not sure why we wouldn't want to do oti and taxomachine as well
19:13 codiferous if we rename everything but don't make any db changes, then backwards compatibility is very easy: we just leave the old classes there with the old names
19:14 codiferous moving forward, i don't think that's the best strategy for maintaining compatibility with old versions of the api, but in this case, it's very low overhead and works
19:17 josephwb of course, all of them
19:17 josephwb i just meant what I was planning on ding myself
19:17 josephwb dpoing
19:17 josephwb doing
19:18 codiferous ah, ok. yeah i had planned to do taxomachine/oti
19:19 josephwb good
19:26 josephwb i wonder: should we separate arguson functions from the others?
19:26 josephwb in treemachine, i mean
19:26 josephwb will anyone else use arguson?
19:27 kcranstn we should eventually phase that out completely
19:27 towodo At least one hackathon participant was interested in node-associated metadata (ncbi & gbif ids)
19:27 towodo currently arguson is the only convenient way to get that
19:27 codiferous i think it's fine to leave arugson in as a format option in the current services
19:27 towodo although people could look up the ott ids in a service that provides that info
19:28 codiferous there's not much advantage to making dedicated services for arguson, but would create duplicate code
19:29 codiferous i do think we should discourage hackathoners from using arguson
19:29 codiferous so maybe breaking them out into services and saying "don't use these" is cleaner than saying "don't use this format" for existing services, idk
19:30 codiferous if they need metadata currently only available via arguson, we can probably make services to provide that pretty quickly during the hackathon
19:30 codiferous but maybe that is overly optimistic
19:31 josephwb right, okay
19:58 towodo here’s another api to look at for inspiration: https://developer.github.com/v3/
22:48 josephwb joined #opentreeoflife
23:52 josephwb joined #opentreeoflife

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