GC log formats
jon.masamitsu at oracle.com
Wed Sep 25 06:47:07 UTC 2013
There were fixes applied to jdk8 for the same (or at
least similar looking) white space bugs.
On 9/24/2013 2:43 AM, Thomas Schatzl wrote:
> Hi Kirk,
> On Sun, 2013-09-22 at 07:10 -0700, Kirk Pepperdine wrote:
>> Someone has introduced changes in the GC log formats in 1.7.0_40. Aside from
>> the dropped "s" on "secs" in some records, the completely missing
>> "secs" on others, we now have a change in the appearance or
>> disappearance of while spaces in the logs.
>> 36.619: [GC 36.619: [ParNew
>> is now
>> 247.165: [GC247.165: [ParNew
>> Arbitrary changes like this one break tools that are out in the ecosystem. In
>> the last short period of time I've lost the ability to use a number of
>> free but closed source tools because of changes in log formats.
>> My question is; are there any tests being run for this type of output so that
>> these seemingly inconsequential formatting changes can be caught?
> I do not think that we are actively trying to break things like that,
> especially if they worsen (human) readability.
> (Please let's not start discussing gc log formatting in this thread :)
> While tests trying to verify log output messages are created as we go,
> they are far from complete.
>> And, can I have my space back please? (and an 's' after sec would be nice also ;-)).
> Could you report an issue on bugs.openjdk.java.net with appropriate
> information? :)
> Patches with a fix are of course highly appreciated.
More information about the hotspot-gc-dev