Refactor ResampleGrid names and type parameter #3121
Conversation
We want to avoid type parameters from being propagated throughout the library when they aren't absolutely necessary. This commit hides the [N: Integral] type parameter which was formerly on each ResampleGrid class
| * | ||
| * @note If used as target of resample operation it acts as identity operation. | ||
| */ | ||
| case object DefaultResampleTarget extends ResampleTarget { |
There was a problem hiding this comment.
DefaultResampleTarget is a bit of a mouthful and none of the other subclasses of ResampleTarget have Resample in them. Might I suggest DefaultTarget as alternative ?
There was a problem hiding this comment.
DefaultTarget sounds much better, but the word Target still sounds not that good as "Grid".
There was a problem hiding this comment.
All changes make sense, perfect that there is no type parameter anymore + very useful comments were added to describe ResampleTarget AST.
However the word ResampleTarget, in particular Target here sounds a bit generic for me (like Data in classes names). I don't have better suggestions though.
| @@ -62,7 +62,7 @@ class GDALWarpOptionsSpec extends FunSpec with RasterMatchers with GivenWhenThen | |||
| rasterExtent = GridExtent(Extent(630000.0, 215000.0, 645000.0, 228500.0), 10, 10), | |||
| CRS.fromString("+proj=lcc +lat_1=36.16666666666666 +lat_2=34.33333333333334 +lat_0=33.75 +lon_0=-79 +x_0=609601.22 +y_0=0 +datum=NAD83 +units=m +no_defs "), | |||
| WebMercator, | |||
| TargetCellSize[Long](CellSize(10, 10)) | |||
| TargetCellSize(CellSize(10, 10)) | |||
| * | ||
| * @note If used as target of resample operation it acts as identity operation. | ||
| */ | ||
| case object DefaultResampleTarget extends ResampleTarget { |
There was a problem hiding this comment.
DefaultTarget sounds much better, but the word Target still sounds not that good as "Grid".
| import spire.implicits._ | ||
|
|
||
| /** Represents a strategy/target for resampling */ | ||
| sealed trait ResampleTarget { |
There was a problem hiding this comment.
Mb these are my language problems, but ResampleGird is a bit more understandable for me rather than ResampleTarget. Is not Target a too generic thing? Like Data when we were designing Paths / DataPaths for RasterSources.
I am not defending ResampleGrid name, I am not fan of it either.
There was a problem hiding this comment.
Target might be too generic, but I guess i'm still a bit stuck on the fact that it isn't a ResampleGrid so much as a strategy/target by which a grid will be summoned/constructed
There was a problem hiding this comment.
@moradology totally agree, it is definitely not a ResampleGrid, I don't have a better naming suggestion, so mb it is fine as you did.
|
LGTM in general, there are only minor naming issues and it can be merged once all comments are answered and CI is happy. |
moradology
force-pushed
the
feature/resample-target
branch
from
6ffd48d
to
ad2c2f4
Compare
* Rename ResampleGrid; hide its type param We want to avoid type parameters from being propagated throughout the library when they aren't absolutely necessary. This commit hides the [N: Integral] type parameter which was formerly on each ResampleGrid class * Rename DefaultResampleTarget to DefaultTarget * Rename TargetGrid to TargetAlignment
Overview
We want to avoid type parameters from being propagated throughout the library when they aren't absolutely necessary. This PR hides the [N: Integral] type parameter which was formerly on
each ResampleGrid class. Additionally, it renames
ResampleGridtoResampleTarget(as the grid is what is produced rather than the case class itself) and renamesDimensionsasTargetDimensionsfor consistency and renames theIdentityResampleGridasDefaultTargetto better reflect its context-dependent nature (the exact operation which is "default" is dependent on which resample/reprojection function is being used.Checklist
docs/CHANGELOG.rstupdated, if necessarydocsguides update, if necessary