Le mar. 26 avr. 2022 à 00:44, David Blevins <[email protected]> a écrit :
> Since we're now getting the snippet max length via the config, could we > construct the JsonGeneratorImpl directly and pass in a buffer of the > correct size? > >From the JsonProvider we can - mapper must not depend on johnzon-core - at the condition to promote the other properties too AND not do it if user provided a custom generator factory which would just be reused in this case, so not sure it solves the issue. > -David > > > > On Apr 25, 2022, at 2:19 PM, Romain Manni-Bucau <[email protected]> > wrote: > > > > No worries, got it. > > > > One issue copying the generator is that it will still miss the flush for > > all small objects and for all other cases the heuristic + a simple > > truncation if needed would work with almost no other downsides than not > > being exact but goal was to limit the overflow/cpu/mem so all good IMHO. > > > > Le lun. 25 avr. 2022 à 22:57, David Blevins <[email protected]> a > > écrit : > > > >>> On Apr 25, 2022, at 10:47 AM, David Blevins <[email protected]> > >> wrote: > >>> > >>> The trick with that is there is yet another buffer being held > internally > >> by ObjectOutputStream and it recreates the issue. > >> > >> Just read this again -- this was supposed to be "OutputStreamWriter" not > >> ObjectOutputStream. Too many years of EJB :) > >> > >> > >> -David > >> > >> > >
