[jira] [Issue Comment Deleted] (COMPRESS-277) IOUtils.skip does not work as advertised

["BELUGA BEHR (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/COMPRESS-277?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

BELUGA BEHR updated COMPRESS-277:
---------------------------------

    Comment: was deleted

(was: Just my 2 cents on some changes...

Yes.  I helped revamp this code quite a bit, but you get what you pay for :)

InputStream#skip declares:

{quote}
Skips over and discards n bytes of data from this input stream. The skip method may, for a variety of reasons, end up skipping over some smaller number of bytes, possibly 0.  This may result from any of a number of conditions; reaching end of file before n bytes have been skipped is only one possibility. The actual number of bytes skipped is returned.  If n is negative, no bytes are skipped.
{quote}

I would recommend doing away with the call to the local IOUtils and just call skip directly on the underlying stream.  I'd also amend the JavaDoc to say "end up skipping .. some smaller number of bytes... reaching the end of the current entry."  The stream is not required to make any best-effort.  For your example, there should be a BufferedInputStream between the TarInputStream and the CipherInputStream.  This would put it more in line with all other InputStreams.

If a client wants to skip an entry manually, they would have to call Commons-IO IOUtils#skipFully, IOUtils#skip, etc.

You would then have to modify getNextTarEntry() to call the IOUtils#skip method.


)

> IOUtils.skip does not work as advertised
> ----------------------------------------
>
>                 Key: COMPRESS-277
>                 URL: https://issues.apache.org/jira/browse/COMPRESS-277
>             Project: Commons Compress
>          Issue Type: Bug
>    Affects Versions: 1.8
>            Reporter: Fabian Lange
>             Fix For: 1.9
>
>
> I am trying to feed a TarInputStream from a CipherInputStream.
> It does not work, because IOUtils.skip() does not adhere to the contract it claims in javadoc:
> "     * <p>This method will only skip less than the requested number of
>      * bytes if the end of the input stream has been reached.</p>"
> However it does:
>             long skipped = input.skip(numToSkip);
>             if (skipped == 0) {
>                 break;
>             }
> And the input stream javadoc says:
> "     * This may result from any of a number of conditions; reaching end of file
>      * before <code>n</code> bytes have been skipped is only one possibility."
> In the case of CipherInputStream, it stops at the end of each byte buffer.
> If you check the IOUtils from colleagues at commons-io, they have considered this case in IOUtils.skip() where they use a read to skip through the stream.
> An optimized version could combine trying to skip, then read then trying to skip again.



--
This message was sent by Atlassian JIRA
(v6.2#6252)