Issues fixed in MarkLogic Server 10.0-4

XDMP-ARGTYPE error messages do not truncate the error message and output very long messages

The XDMP-ARGTYPE error messages do not truncate the error message and output entire sequences or map:map instead of applying ellipses in the same manner that other error messages do. Additionally, when applying the simple mapping operator ! and the || string concatenation expression, the (extremely large) map is serialized in the message when in other exceptions it is not.

Monitoring History: XDQP SERVER REQUESTS DETAIL not rendering

the XDQP Server Request Detail page is rendering empty. It is missing the heading with the XDQP SERVER REQUESTS DETAIL heading without the blue arrow icon to link back to the main page, and does not render the XDQP Server Request Rate or XDQP Server Request Time detail panels.

MLCP export with redaction throws XDMP-EXPNTREECACHEFULL

MLCP export query redacts on all the files returned from a cts search result instead of the correct splits, which causes an XDMP-EXPNTREECACHEFULL error

Namespace missing in xdmp:with-namespaces() for selected element children of the root

An upgrade to 10.0-2.1 causes missing namespace in xdmp:with-namespaces()

Ordering a TDE view by a nullable column is slow

Slow response is observed when using Orderby in an optic query on the template with a large set of documents by a nullable column.

Session timeout for Admin UI, Query Console, and Monitoring History/Dashboard

Implemented session timeout feature for Admin UI, Query Console, and Monitoring History/Dashboard is now available. If the session is idle for more than the specified session time out, user will be logged out automatically from the Admin Ui, Query console or Monitoring history

False negatives in index resolution with punctuation sensitive element word query

With element word and word positions enabled, a punctuation sensitive element word query with leading punctuation may fail to match in the index. The query plan shows word query terms starting at a non-zero relative position.

Updating temporal collections LSQT properties required full admin rights to the database

Changing the temporal collection LSQT properties requires full admin rights to the database. Only "admin/temporal" rights should be required.

Creating a Scheduled Task using the ReST API and specifying task-enabled as false results in a task being created as enabled

Making a POST to http://hostname:8002/manage/v2/tasks?group-id=Default where the JSON payload contains "task-enabled" : false results in a Scheduled task being created with task-enabled being set to true.

receive XML properties as a string in a JSON payload

A REST API client can write and read document metadata as JSON or XML. Previously, when using JSON for document metadata, a REST API client could only persist properties in the "http://marklogic.com/xdmp/json/basic" namespace as written and read by the json:transform() function with the json:basic configuration. Now, when writing properties, the client can specify an XML prop:properties fragment as a JSON string. The structure of the JSON fragment is as follows: {"properties":{"$ml.xml": "<prop:properties>...</prop:properties>"}} To ensure that reading the properties as JSON emits the same structure, the properties are persisted with the following marker element: <rapi:json-serialization as="xml"/>

SAML token clearing on xdmp:logout

SAML assertion token is not cleared from cache when user calls xdmp:logout

Segmentation Fault in xdmp::ParseNode::putqs

Segmentation fault occurs in xdmp::ParseNode::putqs during the evaluation of a FLWOR statement.

Deduplication in op.fromTriples() and op:from-triples()

The Optic API op.fromTriples() and op:from-triples() accessors take options to determine whether to remove duplicate triples. Checking for duplicate triples can slow down retrieval of triples even if there are no duplicate triples. Previously, the accessors deduplicated triples by default. Now, the accessors deduplicate only when the option is specified.

datatype lookup for property name with underscore in JSON QBE

When a JSON QBE uses a property with an undeclared datatype in a range query, QBE looks up the datatype of the range index for the property. If the name of the property contains an underscore, the look up uses a name with two underscores and thus the lookup fails with an XDMP-ELEMRIDXNOTFOUND error.

Intermittent wrong results returned for sparql query

Unpredictably incorrect results are returned from a SPARQL query with quads pattern

Backup does not cleanup if one forest finishes backup before the error

If one forest finishes the backup and the other forest got error, the backup directory is not cleanup properly

Host allowed to be removed when there is a scheduled task associated with the host

When scheduled task is configured for a specific host and that if host is removed from a cluster, then the scheduled task is left with entry host field and does not run.

LDAP authentication with digest md5 on RedHat 8

LDAP authentication with digest md5 does not work on Redhat 8

Internal KMS Auditing

Added audit event to record all operations on the internal KMS.

Monitoring History shortcut links do not set correct options

When using the monitoring history 1h, 1d, 30d shortcut links, the granularity option is not producing the expected results.

TDEs return empty or missing results

TDEs return empty or missing results when querying for single nullable columns, but return results when a nonnullable column is added.

Bi-temporal document delete

Document delete not honoured after bi-temporal document is re-inserted

xs:decimal unhandled overflow

Arithmetic with xs:decimal that should result in an overflow error, is returning data.

Repeated request headers get lost in xdmp.httpGet result

Repeated request headers get lost in xdmp.httpGet result. For example: xdmp.httpGet('https://httpbin.org/response-headers?Set-Cookie=a&Set-Cookie=b'); the first set-cookie header gets lost, and only the last one is returned.

"SELECT * from sys_tables" query not listing all views

When a TDE schema has a mix of protected and unprotected views, the query "SELECT * from sys_tables" does not list unprotected views for a user that has access to see the unprotected views .

Restart of MarkLogic on non-bootstrap host is stuck after using mlcmd leave-cluster

Restart of MarkLogic on non-bootstrap host is stuck after the host has been removed from cluster using mlcmd leave-cluster

XDMP-VALIDATEUNEXPECTED thrown for prop:lower-bound

A bad config may be created and you can see XDMP-VALIDATEUNEXPECTED for lower-bound.

Admin UI: Deleting field does not delete corresponding field range indexes

Create a field on a database based on a path. In the same admin UI page, add a string range index on the field. Click ok to create the field. Let the database reindex; Then, delete the field. Look in the "Field Range Indexes" in the admin page for the database and you will see the range index still there, pointing to the now missing field. This prevents subsequent attempts to create the field with a range index.

Admin UI does not honor or retain forests order

A warning message will now be displayed in the Admin UI indicating that you can not use the Admin UI to configure forest order and to do so, you will need to use the Admin API.

Forests stuck in the event of a Journal file failing to close

Under extremely rare circumstances when a write error occurs during the process of closing a forest, the forest can be stuck in a perpetual state of closing - but never completing to reach the closed state.

Tableau extracting a very large dataset is slow and fills up memory on the client

Tableau hangs when performing an extract. MarkLogic is sending the data all at once causing Tableau to fail

sql:timestampdiff('SQL_TSI_SECOND') returns unexpected value.

sql:timestampdiff('SQL_TSI_SECOND',..) result is not correct as it returns the same result as sql:timestampdiff('SQL_TSI_MINUTE',...) which is the difference in minutes

json-property-scope-query gives false negatives

Positions used with non-text JSON value queries may incorrectly miss documents

mlcp with restrict_hosts true and fastload and segment policy may lose data

If you use mlcp with restrict_hosts true and the database has the segment policy, you'll end up with a discrepancy of documents between what mlcp reports that it loaded and what the database contains.

SPARQL with external bindings slow

SPARQL queries under MarkLogic 10 may perform slower than under MarkLogic 8 in some circumstances when passing external bindings into the query.

Audit Trail incomplete for Query Console queries

When auditing is turned on, using events such as "eval" or "document-read" does show a reference to the query in query console but not the contents of the query so some information is missing from the audit log.

GlobalSign Root CA

Added GlobalSign Root CA to MarkLogic TrustStore.

Support for PUT on manage/v2/query-rolesets/{id}/properties has been removed

A typo in the implementation of PUT means that it never worked. On reflection, the semantics of query rolesets are incompatible with the PUT method, so it has been removed.

Monitoring History tooltips on mouseover not working for load panels

The tooltips that appear as you hover over the line charts are broken for numerous panels Database panels on the main page: Lock Wait Load; Lock Hold Load; Deadlock Wait Load

Multi-statement transaction support in ODBC server

Multi-statement transaction support is now available in the ODBC server

large negative xs:long values handled incorrectly

When accessed from a node over XDQP, large negative values may change, or throw XDMP-WRITE error on writing a document, or throw XDMP-CANTDECODEQUERYKIND when searching with rolesets.

Logout

In some situations, Logout functionality fails when two or more MarkLogic applications are open in different tabs of the same browser.

IBM Information Analyzer issuing queries using unsupported function LENGTH()

LENGTH() function is now supported and is the same as calling char_length()

Monitoring Dashboard: Bar chart uses 1000 instead of 1024 when converting from MiB

Disk-space tab bar chart uses 1000 instead of 1024 when converting from MiB. The starting value is MiB reported by the OS, so converting to GB instead of GiB provides results that are inconsistent with values reported by the OS.

GET error on /v1/graphs when graph has triples

The /v1/graphs endpoint of the REST API support the Graphs Protocol for writing and reading graphs of triples. Previously, if a GET request specified a graph with triples that was not written using the REST API, the endpoint responded with a RESTAPI-NODOCUMENT error. Now, if the triples existing, the GET request returns the triples regardless of whether or not the graph was written with the REST API.

mlecat.bat misnomer

On Windows, there is no "mlecat.bat" file. There is instead a "bin" file.

Premature over-optimization of "instance of" on functions that return errors.

If a function is declared with a return type, direct calls to it from "instance of" with the type are prematurely over-optimized. The function doesn't get called at all and the "instance of" expression is simplified to always return true. Errors or side effects that would happen if the function were called don't happen.

FlexRep: View/set verify-cert target in Admin Console

It is not possible to view or set verify-cert target configuration property through the Admin Console

Query Console: Unable to view large files

Query console becomes unresponsive when user tries to view a large file.

If a CF stack is created with a ML license information, the key/licensee cannot be changed through the AdminUI

Changes done to the license key through the admin ui do not have any effect if there is a license key in the user data or /etc/marklogic.conf.

SPARQL query performance degradation in 9.0-11

Some SPARQL queries using VALUES are not optimized correctly

RPM SHA256 signature support

MarkLogic Server rpm packages now include SHA256 signatures both in the header and payload.