Hhvm: apc_cache_info key differences

[zkiiito]

In HHVM, I get different results for apc_cache_info(). The thing really annoys me, is the missing ['cache_list'][0]['info'].

HHVM Version - all

Standalone code, or other way to reproduce the problem

    apc_store('foo', 'bar');
    var_dump(apc_cache_info('user'));

Expected result

array(15) {
  'num_slots' =>
  int(4099)
  'ttl' =>
  int(0)
  'num_hits' =>
  double(0)
  'num_misses' =>
  double(0)
  'num_inserts' =>
  double(1)
  'expunges' =>
  double(0)
  'start_time' =>
  int(1463214762)
  'mem_size' =>
  double(656)
  'num_entries' =>
  int(1)
  'file_upload_progress' =>
  int(1)
  'memory_type' =>
  string(4) "mmap"
  'locking_type' =>
  string(24) "pthread read/write Locks"
  'cache_list' =>
  array(1) {
    [0] =>
    array(10) {
      'info' =>
      string(3) "foo"
      'ttl' =>
      int(0)
      'type' =>
      string(4) "user"
      'num_hits' =>
      double(0)
      'mtime' =>
      int(1463214762)
      'creation_time' =>
      int(1463214762)
      'deletion_time' =>
      int(0)
      'access_time' =>
      int(1463214762)
      'ref_count' =>
      int(0)
      'mem_size' =>
      int(656)
    }
  }
  'deleted_list' =>
  array(0) {
  }
  'slot_distribution' =>
  array(1) {
    [1470] =>
    int(1)
  }
}

Actual result

array(11) {
  ["start_time"]=>
  int(1463214781)
  ["ttl"]=>
  int(-1)
  ["primed_entries"]=>
  int(0)
  ["primed_live_entries"]=>
  int(0)
  ["values_size"]=>
  int(16)
  ["keys_size"]=>
  int(3)
  ["primed_in_file_size"]=>
  int(0)
  ["primed_live_size"]=>
  int(0)
  ["pending_delete_size"]=>
  int(0)
  ["entries"]=>
  int(1)
  ["cache_list"]=>
  array(1) {
    [0]=>
    array(5) {
      ["entry_name"]=>
      string(3) "foo"
      ["in_memory"]=>
      bool(true)
      ["ttl"]=>
      int(0)
      ["mem_size"]=>
      int(16)
      ["type"]=>
      int(1)
    }
  }
}

[zkiiito]

see: #1853 #5505

[Orvid]

As a note, we have the same info in "entry_name" rather than "info".

[zkiiito]

thanks i've patched that in my code, and now i've realized that i'm also missing "creation_time" and "mtime", and they are not in hhvm :(

[Orvid]

I don't think we currently track that, although it might be possible to partially infer them based on the expire time...

[zkiiito]

the creation_time and mtime keys (among others) are still missing, could somebody reopen this?

[apeabody]

@zkiiito - if you have a moment, perhaps you can check if PR #7125 provides creation_time and mtime is the way you expect, documentation on these is rather scarce. Thanks!

[zkiiito]

it looks good, thanks!