Open kares opened 4 years ago
should also use the opportunity to drop the FileWatch::BufferedTokenizer
which is actually part of LS instead of the plugin!
BufferedTokenizer
uses a Java API for a Ruby split
method. the method writes back-ref $~
which means it needs proper framing, otherwise things might fail or overwrite $~ in unexpected ways. a regexp-less line index
-ing would do the job.
when a custom delimiter is used instead of
"\n"
e.g.delimiter => "</Some>"
(assuming "multi-line" content in a single line), the plugin fails to properly split lines - this mostly happens when the plugin manages to buffer up more than one "line":"<Some><Content1>...</Content1></Some> <Some><Content2>...</Content2>"
and boils down to:
which ends not extracting the 2 lines properly.