Species not eliminated using the character Winter bud scales

newfs / gobotany-app

Deployable code for the Go Botany application

9 stars 8 forks source link

Species not eliminated using the character Winter bud scales #385

Closed sidkoul closed 11 years ago

sidkoul commented 11 years ago

All Acers (family Sapindaceae) have the value "perulate" for the character winter_bud_covering. However, using either the Simple or Full Keys, and selecting "naked" for this character does not eliminate the Acers.

To reproduce:

Visit the Woody angiosperms pile on the Dev server at /simple/woody-plants/woody-angiosperms/
Select "get more questions" and choose "buds"
Choose the question "Winter bud scales?" from the list and set the value to "naked"
Observe that of the six Acers displayed, only two are eliminated.

You can inspect the data here (Dev): /edit/cv/woody-angiosperms-characters/winter_bud_covering_wa/

jnga commented 11 years ago

Followed the reproduction steps above against Dev and all Acers were eliminated. I got 101 matching species and the first plant is Ailanthus altissima. Does the problem still happen for you, Sid?

jnga commented 11 years ago

Looking at it further, seems this was a data problem: the data value for that character had not been specified for those plants. I remember our filtering algorithm does not eliminate plants for which a value has not been specified. Rather, it leaves plants in the results for which the database still needs filling in.

On my local instance, where I have not refreshed data in a few days, I can reproduce the problem in both keys, and can see that the values are missing for those characters in the Admin.

On the Dev server, the problem does not appear as of this morning in either key. The data values now appear filled in there in the Admin.

Assigning back to Sid, in case the values still need to be added to the source data.

sidkoul commented 11 years ago

This is odd. Yes, it's working for me as well now. So what changed? Well we have a new data set and a new build.

The data for winter_bud_covering_wa had been filled in Friday morning (on the NSF Windows server), the data successfully exported and uploaded (gobotany-csv-exports-2013-02-22.zip) to S3 Friday evening. When I checked the dev server Monday afternoon, the Acers were not eliminated, despite the data being there. Now they are. Perhaps the build wasn't successful.

I'm going to modify the email script so it generates a meaningful subject heading when it sends out rebuild log file.

Another question: how do database changes made via the admin effect Simple Key operations?

jnga commented 11 years ago

Database changes should show up right away for local development but might be delayed in a production setting because we have more caching of API calls going on there, I think. (Discussion in standup about our caching in progress)

sidkoul commented 11 years ago

Currently cache values are hard coded in gobotany/api/urls.py. We should pull them into settings.py and have those settings overridden via environment variables (I'll open an issue for this).