Re: http stats

2019-04-04 Thread Dk Jack
Thanks Lief, Bryan. Dk. > On Apr 4, 2019, at 10:07 AM, Bryan Call wrote: > > We have never run without stats on a production machine to my knowledge. > Also, I have never seen stats come up as using a lot of CPU while profiling > when running a benchmark. The comment should definitely be

Re: http stats

2019-04-04 Thread Bryan Call
We have never run without stats on a production machine to my knowledge. Also, I have never seen stats come up as using a lot of CPU while profiling when running a benchmark. The comment should definitely be changed. -Bryan > On Apr 3, 2019, at 9:19 PM, Dk Jack wrote: > > Hi, > I see this

Re: TS API change proposal: Reduce performance impact of TSDebug calls when the tag is disabled

2019-04-04 Thread Aaron Canary
When I've done debug tags like this in the past, it ended up like a simpler version of records.config. IE bool your_tag = false; register_tag("your_tag_name", &your_tag); //store a map can be referenced by name. if (your_tag) printf(...); // direct access, no derefences or indirection with addit

Re: http stats

2019-04-04 Thread Leif Hedstrom
> On Apr 3, 2019, at 10:19 PM, Dk Jack wrote: > > Hi, > I see this comment in records.config > > ## > # The HTTP stats are expensive, turn off you don't need them # > ## > C

Re: TS API change proposal: Reduce performance impact of TSDebug calls when the tag is disabled

2019-04-04 Thread Leif Hedstrom
> On Apr 4, 2019, at 8:05 AM, Walt Karas > wrote: > > Because senility. probably "on" should be a non-pointer of type const > volatile char, and InkAPI in the core can just keep a pointer to the whole > structure for updating. > > But I'm leaning more to fixing up the capability that already

Re: http stats

2019-04-04 Thread Leif Hedstrom
> On Apr 3, 2019, at 10:19 PM, Dk Jack wrote: > > Hi, > I see this comment in records.config > > ## > # The HTTP stats are expensive, turn off you don't need them # > ## > C

Re: TS API change proposal: Reduce performance impact of TSDebug calls when the tag is disabled

2019-04-04 Thread Walt Karas
Because senility. probably "on" should be a non-pointer of type const volatile char, and InkAPI in the core can just keep a pointer to the whole structure for updating. But I'm leaning more to fixing up the capability that already kinda there. Which is optimizimg only for the case when all debug o

Re: TS API change proposal: Reduce performance impact of TSDebug calls when the tag is disabled

2019-04-04 Thread Alan Carroll
This could enable a significant speed up for debug tags. One point is that when the debug tag string is set, at that point the debug objects could be updated to the correct state according to the debug tag string, rather than checking it every time a debug message is logged. I have to ask, why is