OpLib caching doesn't survive after initializing of parrot trace.

[bacek]

Hello.

Main problem - debug_core reinitialize core pmcs.

Simple testcase:

.sub "main"
    .local pmc oplib
    oplib = new 'OpLib'
    trace 1
    $P0 = oplib['say_sc']
.end

~/src/parrot (master)$ ./parrot t.pir 
0005 set P0, P1["say_sc"]                                        P0=PMCNULL P1=OpLib=PMC(0x12bf0f0) 
Segmentation fault

-- Bacek

[plobsing]

The testcase appears to work now. Can this ticket be closed?

[jkeenan]

Hmm. I got:

$ cat t.pir

.sub "main"
    .local pmc oplib
    oplib = new 'OpLib'
    trace 1
    $P0 = oplib['say_sc']
.end

$ ./parrot t.pir
OpLib must be initialized with an oplib name
current instr.: 'main' pc 0 (t.pir:6)

Am I doing something wrong? Or is this not really fixed?

kid51

[plobsing]

Replying to jkeenan:

Hmm. I got:

$ ./parrot t.pir
OpLib must be initialized with an oplib name
current instr.: 'main' pc 0 (t.pir:6)

Am I doing something wrong? Or is this not really fixed? kid51

Sorry, I forgot to mention that with the changes to the way dynops work, OpLib has changed slightly. You now have to specify the name of the oplib you want in the constructor. The modified testcase is:

.sub "main"
    .local pmc oplib
    $P0 = box 'core_ops'
    oplib = new 'OpLib', $P0
    trace 1
    $P0 = oplib['say_sc']
.end

[bacek]

Hello.

Since dynoplib_mapping merged OpLib isn't singletone anymore. Resolving ticket.

-- Bacek