You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@commons.apache.org by "John Vines (JIRA)" <ji...@apache.org> on 2013/10/28 19:26:31 UTC

[jira] [Commented] (VFS-487) File creation events missed when deleted then re-created within poll delay

    [ https://issues.apache.org/jira/browse/VFS-487?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13807059#comment-13807059 ] 

John Vines commented on VFS-487:
--------------------------------

What about explicitly removing files from the stack if they get added, to ensure that this condition doesn't occur?

> File creation events missed when deleted then re-created within poll delay
> --------------------------------------------------------------------------
>
>                 Key: VFS-487
>                 URL: https://issues.apache.org/jira/browse/VFS-487
>             Project: Commons VFS
>          Issue Type: Bug
>    Affects Versions: 2.0
>            Reporter: Sam Haldane
>
> To reproduce:
> * create a file monitor and add a directory with at least one child file
> * set delay to 5 seconds and start the monitor
> * delete a child file, wait for monitor to detect deletion
> * re-create the child file within 5 seconds
> * creation event is not detected
> This is mainly due to this block being at the start of the main loop instead of the end:
> {code:java}
>             while (!this.deleteStack.empty())
>             {
>                 this.removeFile(this.deleteStack.pop());
>             }
> {code}
> When this is executed the file is in the stack from the previous poll and the file has already been re-created. {{removeFile}} calls {{FileMonitorAgent#resetChildrenList}} which updates the children list for that agent (which will contain the newly created file), but doesn't update the monitor map. The subsequent call to {{FileMonitorAgent#check}} in the main loop then doesn't detect the new file.
> Moving the above block to the end of the main loop would prevent most occurrences of this issue, but the implementation is still racy.



--
This message was sent by Atlassian JIRA
(v6.1#6144)