Add function to Optionally get site.info if not present
#3324
Conversation
In do_conversion.R Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
|
@meetagrawal09 can you cross check if corresponding changes in |
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
|
I think you're taking the wrong approach to task 2 here: Yes, "if siteID is not provided, construct a unique identifier from lat/lon", but it needs to be constructed without using the DB. As we discussed in Slack, this could be as simple as something like |
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
|
Should we add a unit test to check if |
Co-authored-by: Chris Black <[email protected]>
There was a problem hiding this comment.
Posting my notes from our live review -- they're a bit cryptic, but hopefully enough to remind you of the key points from the conversation.
Basically I think the entire get.new.site function is trying to handle too many cases, and instead when the db isn't available we should simply paste together lat and lon to use as a siteid.
|
|
||
| if (is.null(site$id) | is.na(site$id)) { | ||
| if ((!is.null(site$lat) && !is.null(site$lon)) | | ||
| (!is.na(site$lat) && !is.na(site$lon)) |
There was a problem hiding this comment.
Weird little gotcha here -- the !is.null will fire when lat and lon are NA, and the !is.na will return a missing value when lat and lon are null.
There was a problem hiding this comment.
You probably want && instead of |
base/db/R/get.new.site.R
Outdated
| lat = site$lat, | ||
| lon = site$lon | ||
| ) | ||
| str_ns <- paste0(new.site$lat, "-", new.site$lon) |
There was a problem hiding this comment.
Might want to use a string other than "-" to concatenate here, to avoid confusion between separator characters and negative lon values
There was a problem hiding this comment.
Replaced current logic with a "_" sign
| } | ||
| } | ||
|
|
||
| site.info <- list(new.site = new.site, str_ns = str_ns) |
There was a problem hiding this comment.
Q in live review: why this structure? A: to match what's expected at met.process line 165
There was a problem hiding this comment.
Detailed Explanation : Current structure is followed to match data flow of what information is later on utilised in met.process for other function calls
base/db/R/get.new.site.R
Outdated
| generate_new_siteID <- function() { | ||
| # Generate a random number. Assuming higher order integers to increase randomness in IDs | ||
| random_id <- sample(10000:99999999, 1) | ||
| return(as.numeric(random_id)) | ||
| } |
There was a problem hiding this comment.
We want IDs to be deterministic -- don't need this
There was a problem hiding this comment.
Can you suggest a method to perform so? I am unable to determine one right now :) !!
There was a problem hiding this comment.
One approach I suggest is as follow : Hashing to generate a unique id using lat and lon of the site. This would ensure greater precision upto 8 decimal points.
| site.info <- PEcAn.DB::get.new.site(site, con=con, latlon = latlon) | ||
|
|
||
| # extract new.site and str_ns from site.info | ||
| new.site <- site.info$new.site |
There was a problem hiding this comment.
Hypothesis to confirm before acting: No downstream object uses new.site$id, can remove it here
There was a problem hiding this comment.
new.site$id is later on passed on to download.raw.met.module for further conversions in convert_input function. I guess for that instance we will have to keep it rather than duplicating the code.
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
There was a problem hiding this comment.
From previous comments on #3348, I couldn't resist removing unnecessary changes. I will post a discussion regarding this PR and gather insights from other 'members & contributors' too.
Signed-off-by: Abhinav Pandey <[email protected]>
Signed-off-by: Abhinav Pandey <[email protected]>
Description
This PR aims to refactor site.id and improve str_ns. The end goal is to make the the System independent of DB. Currently, I'm refactoring to create a siteID if not present already. I'll add test cases to check this util function too.
Work still in pending
Some comments from @mdietze :
Motivation and Context
May Fix a Subtask of #3307
Review Time Estimate
Types of changes
Checklist: