[
https://issues.apache.org/jira/browse/CSV-278?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17380009#comment-17380009
]
David Mollitor commented on CSV-278:
------------------------------------
{code:none}
Benchmark Mode Cnt Score Error Units
CSVBenchmark.parseCommonsCSV-CSV277 avgt 20 5893.317 ± 67.596 ms/op
CSVBenchmark.parseCommonsCSV-master avgt 20 6269.505 ± 80.153 ms/op
{code}
> Reuse Buffers in Lexer for Delimiter Detection
> ----------------------------------------------
>
> Key: CSV-278
> URL: https://issues.apache.org/jira/browse/CSV-278
> Project: Commons CSV
> Issue Type: Sub-task
> Reporter: David Mollitor
> Priority: Minor
> Time Spent: 1h 20m
> Remaining Estimate: 0h
>
> When a character is read, and it could be the first character of the
> delimiter, a look-ahead is performed on the stream to determine if this will
> be the delimiter. This look ahead initiates a new buffer to read into for
> each call. This can cause overhead and much more turn in the heap.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
