Reintroduce convenience where appropriate about elasticsearch-net HOT 4 OPEN

Hi @stefanobranco,

1. The aggregation change requires a little bit of additional code, but on the same time this new structure maps closely to the corresponding JSON request. This is a good thing as it's consistent now with the REST documentation and as well allowed me to remove a lot of internal hacks that were previously required to support the old strucutre in the .NET client (these hacks caused several issues in the past).
2. I will bring back the parameterless MatchAll(). This one somehow got lost in the refactoring process 🙂
3. The HighlightField case I have to check in detail, but it seems like you are mixing up descriptors and class initializer style here. A solution would be to initialize a HighlightFieldDescriptor with your default values instead of HighlightField:

private HighlightFieldDescriptor HighlightField = new().PreTags(...).PostTags(...);

.Highlight(h => h.Fields(fs => fs.Add(new Field("field"), HighlightField)))

from elasticsearch-net.

.Order(new List<KeyValuePair<Field, SortOrder>>(new []{new KeyValuePair<Field, SortOrder>(Field.CountField, SortOrder.Desc)})))

You don't need the List, and depending on the C# version you can write it like this:

// C# 9
.Order(new KeyValuePair<Field, SortOrder>[] { new(Field.CountField, SortOrder.Desc) })

// C# 12
.Order([new(Field.CountField, SortOrder.Desc)])

from elasticsearch-net.

Hi @flobernd!

Thanks for the quick response.

1. The aggregation change requires a little bit of additional code, but on the same time this new structure maps closely to the corresponding JSON request. This is a good thing as it's consistent now with the REST documentation and as well allowed me to remove a lot of

I figured it was something like that. I guess I find the 'Add' a bit 'unattractive', but that's probably a personal thing and it's completely reasonable as it is. I did notice one more small thing, as I'm trying to now completely move away from our nest client. When ordering aggregations:

Before (Nest):

.Order(o => o.CountDescending())

Now (unless I missed something):

.Order(new List<KeyValuePair<Field, SortOrder>>(new []{new KeyValuePair<Field, SortOrder>(Field.CountField, SortOrder.Desc)})))

2. I will bring back the parameterless MatchAll(). This one somehow got lost in the refactoring process 🙂

Perfect, I figured this was just an oversight

3. The HighlightField case I have to check in detail, but it seems like you are mixing up descriptors and class initializer style here. A solution would be to initialize a HighlightFieldDescriptor with your default values instead of HighlightField:

You're right. I'm pretty sure the previous variant did work that way but either way your solution works just as well and is more consistent, so there's probably no need to spend more time on this.

from elasticsearch-net.

.Order(new List<KeyValuePair<Field, SortOrder>>(new []{new KeyValuePair<Field, SortOrder>(Field.CountField, SortOrder.Desc)})))

You don't need the List, and depending on the C# version you can write it like this:

// C# 9
.Order(new KeyValuePair<Field, SortOrder>[] { new(Field.CountField, SortOrder.Desc) })

// C# 12
.Order([new(Field.CountField, SortOrder.Desc)])

Hahaha you're right of course, the list is beyond silly here. I guess the C# 12 version is reasonable enough in that case.

from elasticsearch-net.