Hide much infrastructure in data #2300
Conversation
Codecov Report
@@ Coverage Diff @@
## master #2300 +/- ##
==========================================
+ Coverage 71.14% 71.14% +<.01%
==========================================
Files 779 779
Lines 140243 140251 +8
Branches 16024 16024
==========================================
+ Hits 99773 99783 +10
+ Misses 36038 36035 -3
- Partials 4432 4433 +1
|
| @@ -31,7 +31,8 @@ namespace Microsoft.ML.Transforms | |||
| /// It will be used in conjunction with a filter transform to create random | |||
| /// partitions of the data, used in cross validation. | |||
| /// </summary> | |||
| public sealed class GenerateNumberTransform : RowToRowTransformBase | |||
| [BestFriend] | |||
| internal sealed class GenerateNumberTransform : RowToRowTransformBase | |||
There was a problem hiding this comment.
internal [](start = 4, length = 8)
why not public?
There was a problem hiding this comment.
Pursuant to #1995, all IDataTransforms (as opposed to IEstimator/ITransformers) are to be hidden. So when hiding the entry-point exposing this class, I happened to opportunistically hide this as well, while I was at it.
|
Can you please sync to the latest in master because I’m really curious what the new code coverage is? |
If it naturally comes up due to an actual required change I'd happily rebase again to master, as I always do. In the absence of any requested changes, I see no particular urgency. This might not be the best PR if you're curious about changes in code coverage, since I did not change any tests, or any infrastructure really. I just made things internal. If you are curious about this, you can always of course check out my branch yourself and run the coverage numbers locally. In principle it shouldn't change much, since I did not change much infrastructure, just made it internal. (But still according to the report in its current form, I have a +3.29% change in coverage. Who can account for such things.) I've seen lots of strange stuff like that. Are we certain those code coverage numbers are reliable? |
The goggles do nothing! |
| @@ -0,0 +1,137 @@ | |||
| Data.CustomTextLoader Import a dataset from a text file Microsoft.ML.EntryPoints.ImportTextData ImportText Microsoft.ML.EntryPoints.ImportTextData+Input Microsoft.ML.EntryPoints.ImportTextData+Output | |||
There was a problem hiding this comment.
CustomTextLoader [](start = 5, length = 16)
Why we need this file?
There was a problem hiding this comment.
We don't, sorry. 😄 I'll remove it. WinMerge keeps creating these stupid files...
* Parse/TryUnparse methods internalized. * Lockdown of column abstract class constructors to make them unextensible.
|
Did you mean to check this |
| @@ -37,10 +44,13 @@ public enum CachingOptions | |||
| public abstract class LearnerInputBase | |||
| { | |||
| /// <summary> | |||
| /// The data to be used for training. | |||
| /// The data to be used for training. Used only in entry-points, since in the API the expected mechanism is | |||
| /// that the user iwll use the <see cref="IEstimator{TTransformer}.Fit(IDataView)"/> or some other train | |||
There was a problem hiding this comment.
Did you mean to check this
core_ep-list.tsv.bakfile in?
No. @Ivanidzo4ka had the same comment, I think if you refresh the view you're using you should note that it is removed.
There was a problem hiding this comment.
typeo-
iwll
Whoops. Since internal only I will fix in next phase, if that's OK.
ghost
locked as resolved and limited conversation to collaborators
Another of many steps towards #1602. Commits logically structured. No overall theme, just lots of hiding of individual components, most notably command line parsing, entry-point declarations, and other such things.