IBMStreams/streamsx.nlp

Lemmatizer operator memory allocation issue

vitalizoubov opened this issue · 2 comments

In one of my jobs the nMemoryConsumption metric of the PE, in which the Lemmatizer operator is running, is steadily growing up. Looks like the gposttl Tokenizer function in the Lemmatizer_cpp.cgt module allocates the space that doesn’t get freed.

The wrong usage of the gposttl Tokenizer() and tag() function are causing memory leaks.
After fixing this, there still seems to be a problem in the tag() function allocating memory that doesn't get released.

Fixed in toolkit version 1.2.1