[
https://issues.apache.org/jira/browse/GROOVY-7414?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14532417#comment-14532417
]
Paul King edited comment on GROOVY-7414 at 5/7/15 10:42 AM:
------------------------------------------------------------
I agree those methods and the underlying classes aren't perfect. Do you have
some improvements in mind?
Here is a typical use case:
{code}
def outputBuffer = new StringBuffer()
def errorBuffer = new StringBuffer()
zipProcess = 'gzip -c'.execute()
unzipProcess = 'gunzip -c'.execute()
unzipProcess.consumeProcessOutput(outputBuffer, errorBuffer)
zipProcess.consumeProcessErrorStream(errorBuffer)
zipProcess | unzipProcess
zipProcess.withWriter { writer ->
writer << 'Hello World'
}
unzipProcess.waitForOrKill(1000)
println 'Output: ' + outputBuffer // Output: Hello World
println 'Error : ' + errorBuffer
{code}
The {{withWriter}} method forks a new thread and buffers the stream. What would
you prefer to see?
was (Author: paulk):
I agree those methods and the underlying classes aren't perfect. Do you have
some improvements in mind?
Here is a typical use case:
{code}
def outputBuffer = new StringBuffer()
def errorBuffer = new StringBuffer()
zipProcess = 'gzip -c'.execute()
unzipProcess = 'gunzip -c'.execute()
unzipProcess.consumeProcessOutput(outputBuffer, errorBuffer)
zipProcess.consumeProcessErrorStream(errorBuffer)
zipProcess | unzipProcess
zipProcess.withWriter { writer ->
writer << 'Hello World '
}
unzipProcess.waitForOrKill(1000)
println 'Output: ' + outputBuffer
println 'Error : ' + errorBuffer
{code}
The {{withWriter}} method forks a new thread and buffers the stream. What would
you prefer to see?
> Process output not always consumed before waitFor() returns
> -----------------------------------------------------------
>
> Key: GROOVY-7414
> URL: https://issues.apache.org/jira/browse/GROOVY-7414
> Project: Groovy
> Issue Type: Bug
> Components: groovy-jdk
> Affects Versions: 2.4.3
> Environment: Linux junk 3.16.0-4-amd64 #1 SMP Debian 3.16.7-ckt9-2
> (2015-04-13) x86_64 GNU/Linux running in a VirtualBox VM
> Reporter: Sergey Gromov
> Assignee: Guillaume Laforge
>
> Sometimes the {{Process.consumeProcessOutput(Appendable, Appendable)}} and
> {{Process.waitFor()}} combination does not work as expected. More
> specifically, sometimes {{waitFor()}} returns before all input is actually
> written into {{Appendable}}.
> The following script reproduces the problem:
> {code:title=bug.gy}
> 100.times {
> def out = new StringBuilder()
> def err = new StringBuilder()
> def p = 'echo stuff'.execute()
> p.consumeProcessOutput(out, err)
> p.waitFor()
> def s = out.toString()
> if (s != 'stuff\n') {
> println 'Bug!'
> println 'OUT: ' + s.inspect()
> }
> }
> {code}
> The ability to reproduce this issue seems to depend on the system
> performance. E.g. I can easily reproduce it in my virtual Linux box. The
> above script fails 2 or 3 times on every launch. OTOH it's quite problematic
> to reproduce on my host Windows machine even though I increased the number of
> tries to 1000. Nevertheless I have reproduced it on Windows, too, a couple of
> times.
> I think this is major because I'm not aware of any viable workaround. The
> only thing that works is to sleep a bit after waitFor() which is often
> unacceptable, and it's impossible to guess for how long to sleep.
> Note that {{Process.consumeProcessOutput(OutputStream, OutputStream)}}
> exhibits the same broken behavior.
> Also note that {{Process.waitForProcessOutput(arg, arg)}} methods work
> flawlessly.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
