Re^4: Really Odd Bug With LWP and Device::SerialPort

[kschwab]

Going down that line, the bug isn't triggered when pulling from an http:// url, but it IS triggered with an https:// (ssl) url. The output from Devel::Peek::Dump looks pretty much identical. Which I suppose means the answer isn't anywhere in MAGIC or utf8. Thanks for the pointers though.

# Devel::Peek::Dump output with https url (triggers bug)
SV = PVMG(0x2fbcc3c) at 0x25e7ff4
  REFCNT = 1
  FLAGS = (PADMY,SMG,POK,pPOK,UTF8)
  IV = 0
  NV = 0
  PV = 0x2ff7034 "<!DOCT"\0 [UTF8 "<!DOCT"]
  CUR = 6
  LEN = 12
  MAGIC = 0x2f9066c
    MG_VIRTUAL = &PL_vtbl_mglob
    MG_TYPE = PERL_MAGIC_regex_global(g)
    MG_LEN = -1

# Devel::Peek::Dump output with http url (doesn't trigger bug)
SV = PVMG(0x2e25434) at 0x2567ff4
  REFCNT = 1
  FLAGS = (PADMY,SMG,POK,pPOK,UTF8)
  IV = 0
  NV = 0
  PV = 0x2fe7c5c "<!DOCT"\0 [UTF8 "<!DOCT"]
  CUR = 6
  LEN = 12
  MAGIC = 0x2d96af4
    MG_VIRTUAL = &PL_vtbl_mglob
    MG_TYPE = PERL_MAGIC_regex_global(g)
    MG_LEN = -1

[runrig]

Not sure I understand...the "MAGIC" doesn't appear (at least for me) until after the substr (when the bug is not triggered), and so I was postulating that maybe it's the lack of MAGIC that triggers the bug. I can't comment on what actually triggers the bug since I'm not using serial ports...

[kschwab]

I think we're on the same page. Basically, I can trigger the bug on strings both with, and without magic. Deep and mysterious enough that I think I'll give up and just keep my fix in place.