You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-issues@hadoop.apache.org by "Gregory Farnum (JIRA)" <ji...@apache.org> on 2009/11/02 23:10:59 UTC
[jira] Created: (HADOOP-6355) Make FileSystemContractBaseTest,
LocalFileSystem, and FileSystem javadoc compliant
Make FileSystemContractBaseTest, LocalFileSystem, and FileSystem javadoc compliant
-----------------------------------------------------------------------------------
Key: HADOOP-6355
URL: https://issues.apache.org/jira/browse/HADOOP-6355
Project: Hadoop Common
Issue Type: Bug
Components: fs
Reporter: Gregory Farnum
Priority: Minor
These do not all seem to play nicely together. ie, plugging the LocalFileSystem into FileSystemContractBaseTest (via a subclass that just overrides setUp to set fs=LocalFileSystem) results in 5 failures and 19 errors out of 28 tests.
Additionally, the unit tests expect a few undocumented behaviors (ie, it expects to get an IOException on mkdirs(subdir/of/a/file) rather than just returning false) and some contradictory behaviors (it expects that delete(directory, false) should succeed on an empty directory) compared to FileSystem's javadoc.
I suspect that most of the fixes should come from changing the unit tests to be more sensible rather than modifying [Local]FileSystem, but however it's done these should all agree on expected behavior!
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-6355) Make FileSystemContractBaseTest,
LocalFileSystem, and FileSystem javadoc compliant
Posted by "Gregory Farnum (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-6355?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12772730#action_12772730 ]
Gregory Farnum commented on HADOOP-6355:
----------------------------------------
I'm not very familiar with how issue tracking is handled here, so very possibly. My own interest is mostly in matching the documentation and the unit tests, though -- that's how I discovered the differences.
> Make FileSystemContractBaseTest, LocalFileSystem, and FileSystem javadoc compliant
> -----------------------------------------------------------------------------------
>
> Key: HADOOP-6355
> URL: https://issues.apache.org/jira/browse/HADOOP-6355
> Project: Hadoop Common
> Issue Type: Bug
> Components: fs
> Reporter: Gregory Farnum
> Priority: Minor
>
> These do not all seem to play nicely together. ie, plugging the LocalFileSystem into FileSystemContractBaseTest (via a subclass that just overrides setUp to set fs=LocalFileSystem) results in 5 failures and 19 errors out of 28 tests.
> Additionally, the unit tests expect a few undocumented behaviors (ie, it expects to get an IOException on mkdirs(subdir/of/a/file) rather than just returning false) and some contradictory behaviors (it expects that delete(directory, false) should succeed on an empty directory) compared to FileSystem's javadoc.
> I suspect that most of the fixes should come from changing the unit tests to be more sensible rather than modifying [Local]FileSystem, but however it's done these should all agree on expected behavior!
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Resolved: (HADOOP-6355) Make FileSystemContractBaseTest,
LocalFileSystem, and FileSystem javadoc compliant
Posted by "Jakob Homan (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-6355?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jakob Homan resolved HADOOP-6355.
---------------------------------
Resolution: Duplicate
I'm going to go ahead and say this does duplicate the other jira. It would certainly be great to iron out those differences, but we ran into problems involving backwards compatibility when we worked on it previously. If possible, stick to the FileSystemContract test and then rely on DFS behavior as guidance if there are differences.
> Make FileSystemContractBaseTest, LocalFileSystem, and FileSystem javadoc compliant
> -----------------------------------------------------------------------------------
>
> Key: HADOOP-6355
> URL: https://issues.apache.org/jira/browse/HADOOP-6355
> Project: Hadoop Common
> Issue Type: Bug
> Components: fs
> Reporter: Gregory Farnum
> Priority: Minor
>
> These do not all seem to play nicely together. ie, plugging the LocalFileSystem into FileSystemContractBaseTest (via a subclass that just overrides setUp to set fs=LocalFileSystem) results in 5 failures and 19 errors out of 28 tests.
> Additionally, the unit tests expect a few undocumented behaviors (ie, it expects to get an IOException on mkdirs(subdir/of/a/file) rather than just returning false) and some contradictory behaviors (it expects that delete(directory, false) should succeed on an empty directory) compared to FileSystem's javadoc.
> I suspect that most of the fixes should come from changing the unit tests to be more sensible rather than modifying [Local]FileSystem, but however it's done these should all agree on expected behavior!
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-6355) Make FileSystemContractBaseTest,
LocalFileSystem, and FileSystem javadoc compliant
Posted by "Jakob Homan (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-6355?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12772726#action_12772726 ]
Jakob Homan commented on HADOOP-6355:
-------------------------------------
Is this a duplicate of HDFS-303, which details and encompasses the differences between file system implementations?
> Make FileSystemContractBaseTest, LocalFileSystem, and FileSystem javadoc compliant
> -----------------------------------------------------------------------------------
>
> Key: HADOOP-6355
> URL: https://issues.apache.org/jira/browse/HADOOP-6355
> Project: Hadoop Common
> Issue Type: Bug
> Components: fs
> Reporter: Gregory Farnum
> Priority: Minor
>
> These do not all seem to play nicely together. ie, plugging the LocalFileSystem into FileSystemContractBaseTest (via a subclass that just overrides setUp to set fs=LocalFileSystem) results in 5 failures and 19 errors out of 28 tests.
> Additionally, the unit tests expect a few undocumented behaviors (ie, it expects to get an IOException on mkdirs(subdir/of/a/file) rather than just returning false) and some contradictory behaviors (it expects that delete(directory, false) should succeed on an empty directory) compared to FileSystem's javadoc.
> I suspect that most of the fixes should come from changing the unit tests to be more sensible rather than modifying [Local]FileSystem, but however it's done these should all agree on expected behavior!
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.