Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

[CVS ci] trace output change

3 views
Skip to first unread message

Leopold Toetsch

unread,
Mar 9, 2004, 9:55:35 AM3/9/04
to P6I
I've rearranged the output of parrot -t a bit. It should be more
readable now e.g.:

0 set P16, P1 - P16=NULL, P1=NULL
3 set P17, P5 - P17=SArray=PMC(0x40304908),
P5=SArray=PMC(0x40304908)
6 set I16, P17 - , P17=SArray=PMC(0x40304908)
9 newsub P16, 46, 5715 - P16=Sub=PMC(0x403048f0), ,
13 set I0, 1
16 set I1, 0
19 set I2, 0
22 set I3, 0
25 set I4, 0
28 set P0, P16 - P0=Sub=PMC(0x403048f0), P16=Sub=PMC(0x403048f0)
31 savetop
32 invokecc

The trace_op() function is now called *after* the instruction was
executed, so LHS effects like that of C<new> are immediately visible.

Have fun,
leo

Jens Rieks

unread,
Mar 9, 2004, 10:30:33 AM3/9/04
to perl6-i...@perl.org, Leopold Toetsch
Hi,

Well done, looks really better!

I noticed that DOD runs are disturbing the output:
495 null S31 -
497 set P5, P6 - P5=# DOD run
Data::Dumper=PMC(0x403852c8), P6=Data::Dumper=PMC(0x403852c8)
500 set P6, P7 - P6=NULL, P7=NULL

Good work!

> Have fun,
> leo
jens

Jens Rieks

unread,
Mar 9, 2004, 11:13:21 AM3/9/04
to perl6-i...@perl.org, Leopold Toetsch
Hi again,

I just found another problem related to bytecode switching:

*** switching to BYTECODE_library/Data/Dumper/Base.imc
PC=-4501; OP=<err>

The first shown OP after a switch is always wrong.

jens

Leopold Toetsch

unread,
Mar 9, 2004, 11:08:20 AM3/9/04
to Jens Rieks, perl6-i...@perl.org
Jens Rieks <par...@jensbeimsurfen.de> wrote:

> I noticed that DOD runs are disturbing the output:

I haven't removed it, now it looks like:

2193 setprop P16, ".name", P17 - P16=Sub= ...
DOD
2197 branch 14

Here the C<setprop> did cause a DOD run.

leo

Leopold Toetsch

unread,
Mar 9, 2004, 10:58:21 AM3/9/04
to Jens Rieks, perl6-i...@perl.org
Jens Rieks <par...@jensbeimsurfen.de> wrote:
> Well done, looks really better!

Actually while digging hours through traces of EBNF, I wasn't really
happy with the output, so I changed it ;)

> I noticed that DOD runs are disturbing the output:
> 495 null S31 -
> 497 set P5, P6 - P5=# DOD run

Yep. While it's important to know, when a DOD takes place, if you are
searching for DOD bugs, it's disturbing for the normal case. I'll turn
that output off.

> Good work!

Thanks,
leo

Leopold Toetsch

unread,
Mar 9, 2004, 11:46:35 AM3/9/04
to Jens Rieks, perl6-i...@perl.org
Jens Rieks <par...@jensbeimsurfen.de> wrote:
> Hi again,

> I just found another problem related to bytecode switching:

> *** switching to BYTECODE_library/Data/Dumper/Base.imc
> PC=-4501; OP=<err>

Thanks, fixed.

> jens

leo

0 new messages