Hello -cafe! I got some experience profiling memory leaks using -prof and the +RTS -hc -L350 --RTS option. This gives me a .hp file with samples done at regular intervals. I can feed this file to `hp2pretty`, get some nice graphs and when I identify a suspicious band, I can grep the .hp file to get the complete "cost-center stack" responsible for that band. Here is an example of one: (18514)unsafeCreate/take/hexString/withText/parseOptionalFieldWith<...> I have been trying to replicate this setup using the eventlog logging. I use +RTS -l-aug -hc --RTS to get the eventlog with user and GC events. I use `eventlog2html` to get better view of memory allocation per cost-center. The normalized view presents me with cost-centers at the right side where I can filter on them[1]. But those labels are just the "top" of the cost-center stack. In fact I don't think the whole cost-center stack is represented anywhere on the eventlog file. I can sort of get the structure of the cost-center stack from the .hp file. The labels identifying the cost-center are different between the two files. Is there a way to recover the complete cost-center stack purely from the eventlog file? Thanks in advance [1]: https://imgur.com/a/nXyCEpG -- Rubén. (pgp: 1E88 3AC4 89EB FA22)