45 lines
1.9 KiB
Plaintext
45 lines
1.9 KiB
Plaintext
[[release-notes-8.0.0-alpha1]]
|
|
== {es} version 8.0.0-alpha1
|
|
|
|
coming[8.0.0]
|
|
|
|
The changes listed below have been released for the first time in {es}
|
|
8.0.0-alpha1.
|
|
|
|
[[breaking-8.0.0-alpha1]]
|
|
[discrete]
|
|
=== Breaking changes
|
|
|
|
Aggregations::
|
|
* Disallow specifying the same percentile multiple times in percentiles aggregation {es-pull}52257[#52257]
|
|
|
|
Mapping::
|
|
* Dynamic mappings in indices created on 8.0 and later have stricter validation at mapping update time.
|
|
(e.g. incorrect analyzer settings or unknown field types). {es-pull}51233[#51233]
|
|
|
|
Update by query::
|
|
* Unsupported fields provided as part of the script section of the update by query API are now rejected.
|
|
|
|
Deprecations::
|
|
* Remove undocumented endpoints of hot threads API {es-pull}55109[#55109]
|
|
|
|
Slow loggers::
|
|
* `index.indexing.slowlog.level` and `index.search.slowlog.level` are removed. These settings can be worked around
|
|
by using appropriate thresholds. If for instance we want to simulate `index.indexing.slowlog.level` = `INFO` then
|
|
all we need to do is to set `index.indexing.slowlog.threshold.index.debug` and
|
|
`index.indexing.slowlog.threshold.index.trace` to `-1` {es-pull}57591[#57591]
|
|
|
|
Search::
|
|
* Consistent treatment of numeric values for range query on date fields without `format` {es-pull}[#63692]
|
|
|
|
[[enhancement-8.0.0-alpha1]]
|
|
[float]
|
|
=== Enhancements
|
|
|
|
Machine learning::
|
|
* The Windows build platform for the {ml} C++ code now uses Visual Studio 2019 {ml-pull}1352[#1352]
|
|
* The macOS build platform for the {ml} C++ code is now Mojave running Xcode 11.3.1,
|
|
or Ubuntu 20.04 running clang 8 for cross compilation {ml-pull}1429[#1429]
|
|
* The Linux build platform for the {ml} C++ code is now CentOS 7 running gcc 9.3 {ml-pull}1170[#1170]
|
|
* Add a new application for evaluating PyTorch models. The app depends on LibTorch - the C++ front end to PyTorch and performs inference on models stored in the TorchScript format {ml-pull}1902[#1902]
|