code
code
> > I would suggest to use per-descriptor hits_addend to replace the per-request hits_addend. Note that every descriptor is evaluated independent anyway. The new per-descriptor hits_addend is more flexible and...
I basically think this bug won't acutally effect the correctness because all JSON parsers should could handle it correctly except this may flush some more bytes out. But flush a...
cc @isker Thanks for the reporting. I can have a try to address this problem if I get some free time and **it doesn't effect the performance too much.** But...
> You’re right that there are other ways to make what we want work. That’s why I only really asked for a documentation change 🌞. But if you have a...
I have give a (two) try this weekend. And if we want to make the Json formatter to support `omit_empty_values` (even only partly), we need to introduce much complexity in...
> > This needs to be a runtime flag protected change since it is user visible. > > Should we add a runtime flag for this ? IMO, it's a...
> ut of interest why #40946 needs no feature flag? > This is a behavior change too or not? Because It's be treated **as a security bug fix.** Although strictly...