[ 
https://issues.apache.org/jira/browse/TIKA-2782?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16687165#comment-16687165
 ] 

Tim Allison edited comment on TIKA-2782 at 11/14/18 9:43 PM:
-------------------------------------------------------------

I had already tried to protect stdout from child to parent.  The trick is that 
log4j wrote to stdout before we could do the redirect.

The current fix works to siphon off anything written to stdout before we 
redirect the streams.  This handles log4j in {{debug="true"}} mode.  However, 
if something grabbed stdout before we grab it and redirect it to stderr, then 
they can continue to write to that printstream and corrupt communication 
between parent and child.  I copied what we're doing in ForkParser.  If there's 
an actual solution to this, let's reopen and fix this.


was (Author: [email protected]):
The current fix works to siphon off anything written to stdout before we 
redirect the streams.  This handles log4j in {{debug="true"}} mode.  However, 
if something grabbed stdout before we grab it and redirect it to stderr, then 
they can continue to write to that printstream and corrupt communication 
between parent and child.  I copied what we're doing in ForkParser.  If there's 
a reasonable solution to this, let's reopen this.

> Protect IPC via stdout in child process in tika-server in -spawnChild mode
> --------------------------------------------------------------------------
>
>                 Key: TIKA-2782
>                 URL: https://issues.apache.org/jira/browse/TIKA-2782
>             Project: Tika
>          Issue Type: Task
>    Affects Versions: 1.19.1
>            Reporter: Tim Allison
>            Assignee: Tim Allison
>            Priority: Blocker
>             Fix For: 2.0.0, 1.20
>
>




--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to