JEP 271: Unified GC Logging - Second pre-review
bengt.rutisson at oracle.com
Mon Nov 9 09:06:44 UTC 2015
On 2015-11-07 16:58, ecki at zusammenkunft.net wrote:
> Besides the question with the "start" tagged redundancy and the multiline events (which as Kirk brought up can be collapsed), I do like the new layout but I find the timestamps a bit hard to understand by looking at it (yes one can get used to it, but there is room to make it a bit more intuitive). Would it be an option to print the "real start time" at the front (before or after the gc id) so it is clear it is the relative start time of the event? (This is especially usefull if the logger decorates with absolute timestamps, so you have the relative ones in a fixed column next to it. It also helps the parser to extract those fixed parts)
> [11.247s][info ][gc ] GC#265 (11.228s - 11.247s) GC young (G1 Evacuation Pause) 109M->71M(128M) 19.208ms
> Sorry for bringing this up late, but it seems to declutter the gc event specific numbers. (Leaving the duration at the end in this case)
It is not too late to do this type of changes. They are easy to do.
With your suggestion the start and end timestamps end up closer to the
timestamp provided by the decorations, but they end up farther away from
the duration. To me it makes more sense to have the duration and the
start and end times close to each other.
> BTW the whole GC tooling community will love you, Bengt! ;)
I know ;)
More information about the hotspot-gc-dev