Migrate RAQL Queries from Version 3.7 to 3.8

In addition, the underlying mechanism used for storing datasets in an In-Memory Store has been extended. Therefore, search attributes must not be configured explicitly. Instead, the cache must be configured as being searchable and dynamic indexing must be enabled (use <searchable allowDynamicIndexing="true" /> in cache configuration file) and the search attributes must be specified in the <storeto> statement using the searchattributes attribute.

Deprecated or Unsupported 3.7 Functions

Replace with 3.8 Function

concat(String column-or-literal, String column-or-literal)

Use the || operator

"date"(String column)

"date"(String column, String format)

to_date

"day"(Date column)

"day"(String column)

"day"(Long column)

extract_day

decimal(Double column) (Unsupported Function)

decimal(Long column)

decimal(String column)

cast(columnas datatype)

"hour"(Date column)

"hour"(String column)

"hour"(Long column)

extract_hour

length(String column) (Unsupported Function)

char_length or character_length

"minute"(Date column)

"minute"(String column)

"minute"(Long column)

extract_minute

"month"(Date column)

"month"(String column)

"month"(Long column)

"month"(Object column)

extract_month

number(String column) (Unsupported Function)

cast(columnas datatype)

round(Object column)

This function no longer accepts objects and implicitly casts them to numeric values.

Mashups that relied on this implicit casting must update queries to use the following:

round(CAST(Object column AS double))

If no number of decimal places are specified, this function now rounds to zero decimals which may also affect mashup results. Mashups that require two decimal places must update queries to use the following:

round(Number column, "2")

"second"(Date column)

"second"(String column)

"second"(Long column)

extract_second

"string"(String column)

cast(columnas datatype)

substr(String column, Integer begin, Integer end)

substring(String column, Integer start-position, Integer length

Note:

This is not a simple substitution as the semantics of the parameters that identify the characters to extract have different meanings in these functions.

In addition to different parameters, character positions in substr use zero-based indexes while substring uses 1-based indexes.

to_long(String column)

cast(columnas datatype)

"week"(Date column)

"week"(String column)

"week"(Long column)

extract_week

"year"(Date column)

"year"(String column)

"year"(Long column)

"year"(Object column)

extract_year

Deprecated or Unsupported 3.7 Functions	Replace with 3.8 Function
correlation(Number column, Number column)	corr(Number column, Number column)
covariance(Number column, Number column)	covar(Number column, Number column)
denserank()	dense_rank()
firstvalue(Object column)	first_value(Object column)
analytics.kmean_clusters(String column-list, Integer k, Integer iterations, String measure)	analytics.kmeans_clusters(String column1[,String column2,...String columnN]; Integer k, Integer iterations, String measure) This changes the function name (for consistency) and changes the signature of the function. The syntax change supports a variable number of columns as parameters to identify the features for clustering.
analytics.kmeans_observations(String column-list, Integer k, Integer iterations, String measure)	analytics.kmeans_observations(String column1[,String column2,...String columnN]; Integer k, Integer iterations, String measure) This changes the signature of the function to support a variable number of columns as parameters to identify the features for clustering.
lastvalue(Object column)	last_value(Object column)
mean(Number column) (Unsupported Function)	avg(Number column)
rownumber()	row_number()