[Haskell-cafe] space leak?

[Justin Bailey]

Massimiliano,

I had to update your code for it to compile (removed "sequence" from
testpdf'. However, I don't see any significant difference in the
memory profile of either testpdf or testpdf'.

Not sure how you are watching the memory usage, but if you didn't know
the option "+RTS -sstderr" will print out useful memory statistics
when you run your program. E.g.:

  > pdf_test.exe +RTS -sstderr

gives:

  2,157,524,764 bytes allocated in the heap
  246,516,688 bytes copied during GC (scavenged)
  6,086,688 bytes copied during GC (not scavenged)
  45,107,704 bytes maximum residency (8 sample(s))
       4086 collections in generation 0 (  0.61s)
          8 collections in generation 1 (  0.67s)
        129 Mb total memory in use
  INIT  time    0.02s  (  0.00s elapsed)
  MUT   time    5.83s  (  7.48s elapsed)
  GC    time    1.28s  (  1.45s elapsed)
  RP    time    0.00s  (  0.00s elapsed)
  PROF  time    0.00s  (  0.00s elapsed)
  EXIT  time    0.00s  (  0.00s elapsed)
  Total time    7.13s  (  8.94s elapsed)
  %GC time      18.0%  (16.3% elapsed)
  Alloc rate    369,202,098 bytes per MUT second
  Productivity  81.8% of total user, 65.2% of total elapsed

Above you can see 45 MB was the max amount of memory ever in use - and
according to the heap profiling I did it's about constant. I saw the
same results when using "testpdf'".

A few tricks I've learned to reduce space usage:

  * Use strict returns ( return $! ...)
  * foldl' over foldr unless you have to use foldr.
  * Profile, profile, profile - understand who is hanging on to the
memory (+RTS -hc) and how it's being used (+RTS -hb).
  * Use +RTS -p to understand who's doing all the allocations and
where your time is being spent.
  * Approach profiling like a science experiment - make one change,
observe if anything is different, rollback and make another change -
observer the change. Keep notes!

Good luck!

Justin