Object.freeze() and Object.seal() performance consideration

[vsemozhetbyt]

Recently, @bmeurer has written an article about possible performance penalty for frozen and sealed objects. We have two exposed ones: in zlib.js and in util.js. I cannot assess if there may be any penalty there or can we get away another way there, but maybe @nodejs/v8 can clarify. Sorry if I bother in vain.

[TimothyGu]

FWIW neither of the frozen objects is used as a prototype. I think we are in the clear.

[addaleax]

As I understand the article, the core of the issue is that freezing puts objects into dictionary mode. For the zlib one, that is okay, for the util one that does seem like a bad idea.

[silverwind]

I added the one in util so one cannot delete properties and break inspect, but I guess we could as well drop it, as no other parts of node are really protected against delete.

[BridgeAR]

When I refactored the util code I also changed the access to the object and it performs nicely now. I think I actually benchmarked a regular object against the sealed one as well and there was no difference.

I am going to close this as I do not see anything that is to do right now.