PrintCompilation wall-clock timestamps

[Vitaly Davidovich]

Hi Vladimir,

Thanks for the response.  A few minor comments below.

On Mon, Oct 17, 2016 at 4:40 PM, Vladimir Kozlov <vladimir.kozlov at oracle.com
> wrote:

> We have diagnostic flag to redirect output to hotspot_<pid>.log file:
>
> -XX:+UnlockDiagnosticVMOptions -XX:+LogVMOutput
>
Thanks -- good to know.  This can also capture safepoint stats output (i.e.
PrintSafepointStatistics) as well then, which is nice since it won't go to
stdout.

>
> Or file specified by LogFile diagnostic flag.
>
I thought LogFile was only used in conjunction with LogCompilation, but it
looks like I was wrong about that.  Good to know this as well.

In general, it would be nice to have the option of specifying a separate
file for all compiler related output.  This way we could have one file for
the compilers, and another file(s) for all other VM things.  This is what I
thought UL was going to provide (or allow for), but I could be wrong about
that.

>
> About wall clock usage - file RFE.
>
I'm assuming that's for Chris? :)

>
> Thanks,
> Vladimir



>
>
> On 10/17/16 1:11 PM, Christian Thalinger wrote:
>
>>
>> On Oct 12, 2016, at 3:51 AM, Vitaly Davidovich <vitalyd at gmail.com> wrote:
>>>
>>> Hi guys,
>>>
>>> Has any thought been given to output (or allow an option) the
>>> compilation timestamps in wall clock time rather than millis since JVM
>>> start?
>>>
>>
>> I want that too.  You could contribute a patch.
>>
>> Relatedly, how about being able to specify a file to contain that output,
>>> similar to LogFile for LogCompilation?
>>>
>>> Does the Unified Logging work make any of this possible in Java 9?
>>>
>>> Thanks
>>>
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/hotspot-compiler-dev/attachments/20161017/53c0d78e/attachment-0001.html>