[Release] GroupDocs.Search for .NET v26.11 – Faster searches, leaner indexing, secure search network

Dear GroupDocs Customers,

We are pleased to announce the release of GroupDocs.Search for .NET version 26.11. This release makes repeated and simultaneous searches much faster, reduces the memory used by indexing, and secures the search network with node authentication. Indexing in a search network is now 5 to 20 times faster, and the GroupDocs.Search package includes new builds for .NET Framework 4.8, .NET 6.0, .NET 8.0 and .NET 10.0.

Key New Features
○ Reduce the disk space taken by an index at the cost of loading speed (issue SEARCHNET-3729).
○ Authenticate search network nodes with a shared secret key and restrict the master addresses that a slave node accepts (issue SEARCHNET-3730).
○ Recognize text in images of documents extracted in a separate process (issue SEARCHNET-3731).
○ Search an index repository by chunks (issue SEARCHNET-3732).
○ Highlight words found by a custom word splitter, such as words in text without spaces (issue SEARCHNET-3733).

Major Improvements and Fixes
○ Repeated wildcard, regular expression, numeric range and fuzzy searches reuse earlier matches: repeated regex searches are up to 300 times faster, and repeated fuzzy searches 4 to 30 times faster (SEARCHNET-3736).
○ Simultaneous searches of an index share its search threads: the median time of 8 simultaneous searches dropped from 5.6 ms to 0.18 ms (SEARCHNET-3735).
○ Date range search is up to 10 times faster, and up to 59 times faster in search by chunks. Long date ranges now return complete results instead of being cut off at the occurrence limit (SEARCHNET-3734, SEARCHNET-3739).
○ The default alphabet of new indexes now includes Hebrew, Devanagari, Georgian, extended Latin, Greek and Hangul letters and combining diacritical marks, so words such as “Việt” are no longer split into parts (SEARCHNET-3738).
○ Indexing allocates over 4 times less memory and is up to 2 times faster. Retrieving stored document text and merging indexes are 3.5 times faster and allocate 11 and almost 50 times less memory, respectively (SEARCHNET-3740).
○ Indexing in a search network is 5 to 20 times faster and creates far fewer index segments: documents that arrive while a shard is busy are indexed in batches, and the shards of a node work in parallel (SEARCHNET-3742).
○ Registering documents on the master node no longer slows down as their number grows: 20,000 documents are registered in 0.12 s instead of 27.1 s, with 0.8 MB written instead of 7.6 GB (SEARCHNET-3741).
○ Index, IndexRepository, Cancellation, ExtractionHost and SearchNetworkNode implement the standard dispose pattern, and disposing a Cancellation releases its timer (SEARCHNET-3737, SEARCHNET-3743).
○ The GroupDocs.Search NuGet package now includes builds for .NET Framework 4.8, .NET 6.0, .NET 8.0 and .NET 10.0 in addition to .NET Framework 4.6.2 and .NET Standard 2.1, and selects the right one automatically (SEARCHNET-3693, SEARCHNET-3747).
○ Fixed indefinite waits for lost or unresponsive nodes in a search network: the master node detects lost nodes, and the wait for responses can be limited (SEARCHNET-3746).

Action Required When Upgrading
○ Indexes created by version 26.10 open without conversion, but once changed by version 26.11 they can no longer be opened by earlier versions. Upgrade all applications that share an index at the same time.
○ In .NET Framework projects, replace the GroupDocs.Search.NETFramework package with GroupDocs.Search, which now includes the .NET Framework builds.
○ Upgrade all nodes of a search network together and set the same TcpSettings.NetworkKey (at least 16 characters) on every node: nodes with different keys do not connect. Messages are signed but not encrypted.
○ A network node that is silent for 15 seconds is treated as lost, and calls waiting for it throw SearchException. Set TcpSettings.ResponseTimeout to limit how long the master node waits for responses.
○ IndexSettings.SearchThreads is now the total number of threads shared by all simultaneous searches of an index, not the number of threads of each search.
○ Only new indexes get the extended alphabet. To merge a new index with an index created by an earlier version, give the new index the alphabet and the tokenization dictionary of the older index before adding documents.
○ Search by chunks in an index repository goes through the indexes one by one. Continue with IndexRepository.SearchNext while SearchResult.NextChunkSearchToken is not null.
○ For OCR in a separate extraction process, set ExtractionHost.OcrConnector in the extraction host program, built with the same GroupDocs.Search version as the application.

Reference Links
○ Release notes: GroupDocs.Search for .NET 26.11 Release Notes
○ Migration notes: Migration Notes | GroupDocs
○ GitHub examples: GitHub - groupdocs-search/GroupDocs.Search-for-.NET: GroupDocs.Search for .NET examples, plugins and showcase projects · GitHub
○ Documentation: GroupDocs.Search for .NET | GroupDocs
○ NuGet package for .NET and .NET Framework projects: NuGet Gallery | GroupDocs.Search 26.11.0

This update brings faster searches, leaner indexing and a more secure, dependable search network to projects of any size.

We welcome your feedback. Please direct any questions or feature requests to our support team.

Sincerely,
The GroupDocs Team