Alexey

Sure, will update this. Probably I missed something in my analysis, I've based current change on assumption thats ```InitAddr``` and ```FreeAddr``` is pair function and there is always expected that...

@rousskov Actually there is just an ```new-delete-type-mismatch``` was caught with Address Sanitizer. Memory leak was my assumption. Sorry, if it added some unclearance There typical log from ASan for this...

I'll prefer to stick an approach with encapsulation of construction/deleting ```ai_addr```

Changes looks good from my side. Thank you for polishing

I'll check it patch with ASan again and add comment with results little bit later

Im little bit lost track of the conversation. @rousskov @yadij, now I supposed to push my efforts to getting rid of a dynamic allocations for all usages of ```addrinfo``` related...

Got it

Sorry, didnt start yet. I've some medical issues. I'll continue work after ~2 weeks

I've started analysis of things which we discussed above, and I'm afraid, that I could'nt found any "dummy" usages of addrinfo structure, even Icmp* classes use information provided by recvfrom...

Got it, thanks for hints, will proceed analysis then