[
https://issues.apache.org/jira/browse/COMPRESS-595?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17439572#comment-17439572
]
Peter Lee commented on COMPRESS-595:
------------------------------------
Hi [~nervy_protozoan]
Thank you for your reporting.
I made a test with your test case and found that the current implemention could
be passed. I'm testing with the latest code. Maybe you missed something in your
testcase?
> IOUtils.readRange(ReadableByteChannel input, int len) reads more than len
> when input.read() returns < len (ie. there is a partial read)
> ---------------------------------------------------------------------------------------------------------------------------------------
>
> Key: COMPRESS-595
> URL: https://issues.apache.org/jira/browse/COMPRESS-595
> Project: Commons Compress
> Issue Type: Bug
> Affects Versions: 1.21
> Reporter: NP
> Priority: Major
> Attachments: IOUtilsTest.kt
>
> Time Spent: 40m
> Remaining Estimate: 0h
>
> When `input.read(b)` returns `readNow` < `len`, then it means
> `input.read(b)` will need to be called again with the same
> buffer, whose `remaining()` is now the old `remaining()` - `readNow`.
> This way the ReadableByteChannel knows how many bytes are to be
> read in subsequent iterations of the `while (read < len)` loop.
> This is currently not the case, because there is a call to rewind()
> which results in a buffer whose remaining() is reset to `len` if
> `readNow` < `len`.
> I suspect the readRange() method has only been used with channels that never
> do partial reads (such In Memory Byte Channels backed by an array), and hence
> the problem has not been experienced until now.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
