Thanks for the suggestions! I tried duplicating json-stringifier.cc but ran quickly into the fact that it uses v8::internal stuff which isn't exposed in the v8 public API, and therefore not exposed in Node's plugin APIs either. So, I implemented on top of v8::Object::GetPropertyNames() instead. Thus far, my implementation is 3x slower than the built-in one: together v8::Object::GetPropertyNames() and v8::Object::Get() eat up two thirds of the time. The internal implementation is able to use IterateOwnDescriptors(), which seems much faster (and doesn't allocate an array of keys).
-- Philip On Thursday, January 6, 2022 at 6:48:17 AM UTC-8 ja...@mogill.com wrote: > Hi Phillip, > > When faced with a similar situation we were concerned with startup/restore > efficiency of parsing all that JSON every time the program was run, we > concluded a lazy evaluation strategy worked best for us and we decomposed > the object for serialization & storage, later it was reconstituted when > referenced. Your additional step of compression would complicate this, but > the basic idea holds. > > If you really need a monolithic snapshot, your plan of implementing your > own based on existing deep-copy and stringify code is probably your least > surprise-free permanent solution. If you publish it you may discover > you're not the only one reading and writing large JSON objects. > > However, if it is possible to leverage your knowledge of the data to > decompose it for serialization, a small amount of JS to paper over the > decomposition may be all you need. > > -J > > -- -- v8-users mailing list v8-users@googlegroups.com http://groups.google.com/group/v8-users --- You received this message because you are subscribed to the Google Groups "v8-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to v8-users+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/v8-users/cd49ac74-cdf0-4599-a8d5-8a0993f68340n%40googlegroups.com.
