RRF: JEP-271: Unified GC Logging

[kirk.pepperdine at gmail.com]

> On Nov 24, 2015, at 11:51 AM, Bengt Rutisson <bengt.rutisson at oracle.com> wrote:
> 
> 
> Hi Kirk,
> 
> On 2015-11-23 14:26, kirk.pepperdine at gmail.com <mailto:kirk.pepperdine at gmail.com> wrote:
>> Hi Bengt,
>> 
>> I’ve not seen anything egregious as of yet though I’m no where near done.
> 
> Thanks for looking at this change!
> 
>> One comment for ostream.cpp.
>> 
>> -// log_name comes from -XX:LogFile=log_name, -Xloggc:log_name or
>> +// log_name comes from -XX:LogFile=log_name or
>>  // -XX:DumpLoadedClassList=<file_name>
>> 
>> 
>> IME, -Xloggc:log_name is the most common form used. Not stuck to it, just saying…
> 
> Are you saying that it is scary to remove -Xloggc? Yes, I would agree. We've discussed it a bit here. It would be nice to get rid of it since it comes with a bunch of other issues such as log rotation etc. One way to handle the transition would be to map -Xloggc=filename to -Xlog:gc*:filename. It wouldn't be 100% correct and it wouldn't handle the log rotations flags. But maybe it would be a smoother transition for some users.

It’s a small change and a very sensible one. The only thing I’m suggesting is that it’s worth considering this or any change in light of the complexity/change tax that projects will have to pay to move to 9. I see two arguments here, depreciate in 9, remove in 10. Or, just pile it onto the tax. I guess you could argue that since GC logging flags are already changing, the later solution works best in both the short and long term.

Regards,
Kirk
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/hotspot-gc-dev/attachments/20151125/09228c3a/attachment.htm>