[
https://issues.apache.org/jira/browse/HADOOP-7575?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13106568#comment-13106568
]
[email protected] commented on HADOOP-7575:
-------------------------------------------------------
bq. On 2011-09-16 01:43:35, Ravi Prakash wrote:
bq. >
hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/fs/LocalDirAllocator.java,
line 270
bq. > <https://reviews.apache.org/r/1921/diff/1/?file=41412#file41412line270>
bq. >
bq. > Do you really need to get the URI? Doesn't passing the Path to the
File constructor here have the same effect?
Sadly, there is no direct way to do this since our org.apache.hadoop.fs.Path
object has a java.net.URI and is not a subclass of java.net.URI. This is why we
need two steps.
bq. On 2011-09-16 01:43:35, Ravi Prakash wrote:
bq. >
hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/fs/LocalDirAllocator.java,
line 22
bq. > <https://reviews.apache.org/r/1921/diff/1/?file=41412#file41412line22>
bq. >
bq. > No biggie, but could you please import the specific class you are
using? I know the others imported * too, but its a bad practice (
http://stackoverflow.com/questions/147454/why-is-using-a-wild-card-with-a-java-import-statement-bad
)
Thanks for pointing this article out. After looking more closely, I don't even
need to import this since I don't need the definition of URI to pass it
unchanged from toUri to the constructor of File. Removing import of java.net.*
completely.
bq. On 2011-09-16 01:43:35, Ravi Prakash wrote:
bq. >
hadoop-common-project/hadoop-common/src/test/java/org/apache/hadoop/fs/TestLocalDirAllocator.java,
line 81
bq. > <https://reviews.apache.org/r/1921/diff/1/?file=41413#file41413line81>
bq. >
bq. > Can we please move these definitions at top with the others since
its not being used in this method alone but in later methods too? Its just my
opinion.
Reasonable enough. Done.
bq. On 2011-09-16 01:43:35, Ravi Prakash wrote:
bq. >
hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/fs/LocalDirAllocator.java,
line 265
bq. > <https://reviews.apache.org/r/1921/diff/1/?file=41412#file41412line265>
bq. >
bq. > Does localDirs[i] need to be modified as well? i.e. is there
somewhere else in the code where it might trip up?
the issue only occurs when Strings are passed directly as the parameter to
java.io.File. java.io.File doesn't understand schema as part of its
constructor, only org.apache.hadoop.fs.Path does. I have double checked the
code to verify this issue will not occur anywhere else.
- Jonathan
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/1921/#review1919
-----------------------------------------------------------
On 2011-09-15 20:10:18, Jonathan Eagles wrote:
bq.
bq. -----------------------------------------------------------
bq. This is an automatically generated e-mail. To reply, visit:
bq. https://reviews.apache.org/r/1921/
bq. -----------------------------------------------------------
bq.
bq. (Updated 2011-09-15 20:10:18)
bq.
bq.
bq. Review request for Tom Graves, Jeffrey Naisbitt, Robert Evans, and Ravi
Prakash.
bq.
bq.
bq. Summary
bq. -------
bq.
bq. Contexts with configuration path strings using fully qualified paths (e.g.
file:///tmp instead of /tmp) mistakenly creates a directory named 'file:' and
sub-directories in the current local file system working directory.
bq.
bq.
bq. This addresses bug HADOOP-7575.
bq. http://issues.apache.org/jira/browse/HADOOP-7575
bq.
bq.
bq. Diffs
bq. -----
bq.
bq.
hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/fs/LocalDirAllocator.java
71c8235
bq.
hadoop-common-project/hadoop-common/src/test/java/org/apache/hadoop/fs/TestLocalDirAllocator.java
1e22a73
bq.
bq. Diff: https://reviews.apache.org/r/1921/diff
bq.
bq.
bq. Testing
bq. -------
bq.
bq. unit tests included as part of the patch
bq. manual tests that verifies qualified paths don't accidentally create a
directory named file:
bq.
bq.
bq. Thanks,
bq.
bq. Jonathan
bq.
bq.
> Support fully qualified paths as part of LocalDirAllocator
> ----------------------------------------------------------
>
> Key: HADOOP-7575
> URL: https://issues.apache.org/jira/browse/HADOOP-7575
> Project: Hadoop Common
> Issue Type: Bug
> Components: fs
> Affects Versions: 0.23.0
> Reporter: Jonathan Eagles
> Assignee: Jonathan Eagles
> Priority: Minor
> Fix For: 0.23.0
>
> Attachments: HADOOP-7575-trunk-v1.patch, HADOOP-7575-trunk-v3.patch,
> HADOOP-7575-trunk-v4.patch, HADOOP-7575-trunk-v5.patch,
> HADOOP-7575-trunk-v8.patch
>
>
> Contexts with configuration path strings using fully qualified paths (e.g.
> file:///tmp instead of /tmp) mistakenly creates a directory named 'file:' and
> sub-directories in the current local file system working directory.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
