Perl 6 - the future is here, just unevenly distributed

IRC log for #opentreeoflife, 2014-08-20

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

All times shown according to UTC.

Time Nick Message
00:21 josephwb joined #opentreeoflife
06:37 jimallman joined #opentreeoflife
11:06 jimallman joined #opentreeoflife
12:47 jimallman joined #opentreeoflife
13:23 jimallman joined #opentreeoflife
13:54 kcranstn joined #opentreeoflife
13:57 kcranstn joined #opentreeoflife
14:00 kcranstn joined #opentreeoflife
14:15 jimallman joined #opentreeoflife
14:55 josephwb joined #opentreeoflife
14:55 josephwb hey jimallman
14:55 josephwb what is the status of the name mapping?
14:58 josephwb just checked: looks to be the same.
14:58 jimallman hi josephwb. are we hoping to move this to production today?
14:58 josephwb ok, thanks
14:58 josephwb can you ping me?
14:58 jimallman i’m reviewing other recent changes on master, trying to make sure nothing else (esp. treemachine and oti) have changed.
14:59 jimallman don’t want to break things in another exciting way
14:59 josephwb also: does the proposed node query service i mention yesterday sound useful to you? internet here is choppy, so i may have missed your response.
15:00 josephwb kcranstn seemed skeptical
15:10 * jimallman is reviewing that conversation now...
15:13 jimallman i agree with her response that we’re already able to figure out whether a taxon is present in the graph vs. the synthetic tree, but we’re doing it very indirectly by examining responses from other service calls. we don’t have a clear-cut, definitive test for these conditions. do we need one? maybe, if other callers/clients want to query for a taxon.
15:13 jimallman i also agree that the feature we really don’t have now is the “return nearest ancestor in the synthetic tree”
15:14 jimallman i’m not sure either of these are urgently needed. but in the absense of the MRCA-in-synthetic-tree test, we’ll just start offering links to ancestors that may or may not work
15:14 jimallman josephwb: ^
15:17 josephwb joined #opentreeoflife
15:30 kcranstn joined #opentreeoflife
15:38 jimallman josephwb, kcranstn: I’ve added a tentative pull request with all recent changes, with an eye toward pushing these to production: https://github.com/OpenTreeOfLife/opentree/pull/405
15:38 jimallman there’s a list of caveats, though. so if we want to do something today, i might need to cherry-pick just the features you feel are most urgent.
15:39 jimallman clearly one of these is the stricter OTU mapping (ignoring fuzzy matches)… we could also push the “phase 1” versions of feedback on missing/disconnected taxa, though i’d love to get more responses to towodo’s proposed tweaks.
15:40 jimallman fyi, i’ll be on the road (heading back to RTP) this afternoon, and so out of touch. But I can implement any decisions later tonight.
15:40 kcranstn ok -that’s a big pull request. It will take me a little time to go through it
15:41 kcranstn probably not until this afternoon
15:45 jimallman understood. i’ll check in periodically and try to answer questions.
15:47 josephwb the only thing i need immediately is the fuzzy-match removal.
15:48 jimallman gotcha. let me see if this is an easy cherry-pick to Cavia. if so, i can push that pretty quickly.
15:48 jimallman (i believe it’s all javascript, so should be not-scary)
15:48 kcranstn ha. In my mind, all javascript is scary
16:01 jimallman OK, i’ve tested the stricter OTU mapping (just one cherry-picked commit) and it works locally. Here it is on Cavia:  https://github.com/OpenTreeOfLife/opentree/commit/c9e5b4749dfa2bd447234d4dc7e83adc90eefaf4
16:02 jimallman Based on my testing of study pg_2628 (a known problem study), I believe this is safe to push to production. OK?
16:02 kcranstn is there a way for josephwb to test?
16:03 josephwb is it on devtree?
16:05 josephwb *looking at code now*
16:12 josephwb haven't done much (read: anything) with javascript, but this looks wrong: if (false)
16:12 josephwb placeholder code?
16:17 josephwb jimallman: looks ok to me, although it is not clear what happens if multiple exact matches are found. is the first one returned?
16:22 jimallman yes, the first one will be applied automatically. at least it’s exact!
16:23 kcranstn multiple exact = homonym?
16:23 jimallman the “if (false)…” block is code that will be useful when we take the next step, which is to offer a choice of possible matches found (exact first)
16:24 josephwb don't think the multiple hit thing should be a frequent problem. can put off for later.
16:24 jimallman regarding testing: this stricter OTU mapping behavior is on devtree, yes, for example: http://devtree.opentreeoflife.org/curator/study/edit/pg_2628?tab=otu-mapping
16:25 josephwb works as expected
16:25 kcranstn yay!
16:25 josephwb curiously slow, tho...
16:26 jimallman i test here by un-mapping the first three mapped OTUs —Sarcophaga boettcheri, etc. — then clicking ‘Start mapping’. These three should be mapped to their exact match, while all the other unmapped OTUs with weird names will fail to map.
16:26 jimallman yes, it seems we return a set of n results (100?) as soon as they’re found. non-mapping OTUs have to check everything in OT taxonomy, which does slow things down.
16:33 josephwb joined #opentreeoflife
16:33 josephwb *cursing the network*
16:37 josephwb jimallman: doesn't seem to find genuine synonyms
16:38 josephwb e.g. "Sarcophaga bezziana" (currently mapped) is a synonym for "Sarcophaga infantilis"
16:38 josephwb if i unmap it, it is not refound
16:38 josephwb (apologies for being a headache)
16:39 josephwb i don't know enough about javascript to know why it is so slow (~5 seconds / name)
16:39 josephwb is it sorting?
16:40 josephwb i mean, is that the bottleneck?
16:41 josephwb jimallman ^
16:42 josephwb am i coming through? network is dodgy
16:43 josephwb exact matches are quite fast. failed mappings are quite slow.
16:58 josephwb i tried the "Sarcophaga bezziana" query through curl and get no hit from either api or devapi, so likely a taxomachine problem?
16:58 josephwb was using "autocompleteBoxQuery"
16:59 kcranstn jimallman said he was going to be out of the office this afternoon
16:59 kcranstn thanks for testing this, though. I am glad we didn’t push to production too quickly.
17:04 josephwb seems like there are *no* synonyms in the taxomachine DB? cody is looking at it now.
17:06 josephwb are you still there jimallman?
17:06 kcranstn jimallman said he was going to be out of the office this afternoon
17:06 josephwb ok. missing messages because network keeps dropping me.
17:14 josephwb seems like the fuzzy matching is exceedingly slow
17:14 josephwb seems like synonyms are indeed in the taxomachine DB, but there is a bug somewhere. cody is fixing it
17:19 jimallman josephwb: kcranstn: sorry, i’m back for a few minutes. catching up now...
17:20 jimallman regarding slowness:  non-mapping OTUs have to check everything in OT taxonomy, which does slow things down.
17:20 josephwb yes, cody and i are discussing it now
17:21 josephwb shouldn't check *everything* tho. you mean everything within the context?
17:21 jimallman re: synonyms, there are two flags that *might* make a difference? ask Cody
17:21 jimallman includes_deprecated_ids: false
17:21 jimallman includes_dubious_names: false
17:21 jimallman yes, within the context (default is All Life)
17:21 josephwb jimallman: there is a bug. cody just found it. discussing solution.
17:21 jimallman sweet
17:23 jimallman i’lll need to bail in the next 15 minutes, but i’ll be back home and ready for work tonight.
17:25 jimallman while you’ve got Cody, ask if he can see the cause of false positives in the duplicate-study logic. if we can clear that up, we could push a lot of good stuff to production.
17:32 josephwb ok, will do
17:33 josephwb seems to be a fuzzy matching problem
17:33 jimallman yes, i could see similarities in the matches, but no clear pattern
17:34 jimallman i’ve gotta run here, will check in when i get home (evening). thanks again!
17:34 josephwb me too
17:34 josephwb TNRS is more pressing, so we are working on that for now
17:42 josephwb joined #opentreeoflife
17:59 josephwb kcranstn: cody wants to push updated taxomachine to prod but doesn't know how
18:00 josephwb doesn't seem to be correct config file in deploy
18:00 kcranstn only jim and jonathan have permission to post to production
18:02 josephwb not you too?
18:02 kcranstn I have permission but not knowledge ;)
18:02 josephwb seems like it is quite out of date
18:02 josephwb i will email jonathan then
20:38 kcranstn joined #opentreeoflife
20:50 josephwb joined #opentreeoflife
21:16 josephwb hello world
21:27 josephwb joined #opentreeoflife
22:05 josephwb joined #opentreeoflife

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