globalnamesarchitecture / gnlist-resolver-gui Goto Github PK

Currently we cannot have more than one pod, because non-commercial version of ingress can send request to any pod in the system. We need a organize a mount that can be shared by all pods. This will also allow us to do rolling restart. I also will double check that minikube works correctly.

Use this data in the name of the file:

name of the original file
name of the data source
date for the data source release
date when the match was done

As another ticket -- add a setting that will allow to configure file name.

Add a field that will tell how many rows correspond to this match

We need to figure out feasibility of this approach. If people install the project as a web page they would need to download Elm's JS file from a remote link and then some combination of CORS/Proxy would do the trick of connecting to the server.

This ticket is about an investigation phase. If we find the approach feasible we will set our system to support "application as a static web page" approach.

Placeholder for MDL framework, with initial implentation

Error handling (top level) placeholder

We moved the prototype code from gn_crossmap_web with whole history. As a result when we create a ticket in old repository the numbers are duplicated with old code. We need to flatten old history.

Changes from

to

This is a split from #7 issue, as this work need to be done before switching to gnindex-api resolver. We will have the branch for #7 separate from master for quite a while, and it is good to keep it as similar as possible with master, to make upstream rebases simpler

Infraspecies without authors (Natrix natrix cypriaca) gives n/a as inputRank.

Orobanche densiflora Salzmann ex Reuter in DC. gives inputRank n/a. Why?

see input and result files attached

Infraspecies is written with capital produces wrong canonical match. E.g. Iris humilis Georgi
subsp. Arenaria (Waldst. et Kit.) A.et D.Löve
Astragalus macrocarpus DC. subsp. Lefkarensis

Placeholder for i18n

Ingestion step depends on biodiversity gem. It does make sense to migrate to gnresolver project as it is significantly faster, and has more features. For this we need to move gn_list_resolver gem to JRuby

Output in UTF-8. Why do some accepted species names have special characters depicted
correctly, while others don’t. E.g. & versus & etc.?

If I have data that will not does not have enough information for resolution process, I want to see an error message that explains what needs to be done for this step to succeed. I should not be able to click on "Continue" button in this case.

Migrate to GraphQL API of gnindex

Clone gn_crossmap_web and start refactoring

Simple results have 3 categories:

match
fuzzy
no match

Advanced have
exact + canonical = match
fuzzy = fuzzy
partial + partial fuzzy + genus + unmatch + error = no match

This ticket grew from a bug fix to an improvement, so I'll give estimate it

would expect that Adenophora lilifolia gets some score because Adenophora liliifolia with only one additional character is an accepted name? http://webservice.catalogueoflife.org/col/webservice?name=Adenophora+liliifolia

We do have Material design framework, but it is not polished yet. So we need to enhance it.

We have to normalize headers to the terms that the program understands. To do that we need to create workflow that prevents entering wrong combinations of data

Currently Resolution "page" has to be opened from the start to the end of the process. Users should be able to leave the page, return to it again while the resolution continues to happen in the background

The tax_id field that the resolver imports from CoL and outputs in the results is useless, it is just a database index id, changes with every edition and is not used in other CoL services. Unfortunately CoL is in a transition phase from offering LSIDs earlier to 'natural keys' after AC2017 edition. These are however already in DwC, in the references field. I would suggest to extract these from the references field. Example: from http://www.catalogueoflife.org/annual-checklist/details/species/id/1d761fa6e15f9ba277ad7784af78c8b4/synonym/5fc5c8ab89caeede5ede03be346369a7 the identifier can be extracted for both the synonym and its current accepted name. It may also be helpful to give this whole url in the output.

Upload of the file happens in about the same time, but the 'wait' period after upload increased from 7 to 40 seconds for 1 million names file according to my measurements on production servers

We need to setup kubernetes configurations for both projects

Canonical field shows differences in fuzzy matching, so edit distance next to it is convenient

Napkin design

Simple results have 3 categories:

match
fuzzy
no match

Advanced have
exact + canonical = match
fuzzy = fuzzy
partial + partial fuzzy + genus + unmatch + error = no match

Currently Dockerfile is huge and contains a lot of stuff that is only important for development. We need a new Dockerfile that is shaves off all the unnecessary baggage.