[fpc-pascal] Fingerprinting operation of a program

Mark Morgan Lloyd markMLl.fpc-pascal at telemetry.co.uk
Sat Feb 19 18:43:42 CET 2011


Tomas Hajny wrote:
> On Sat, February 19, 2011 15:18, Mark Morgan Lloyd wrote:
>> Henry Vermaak wrote:
>>> On 18 February 2011 18:19, Mark Morgan Lloyd
>>> <markMLl.fpc-pascal at telemetry.co.uk> wrote:
>>>> I need to trace through a program, running on two different systems
>>>> which
>>>> might or might not have the same OS, in order to find out where it's
>>>> failing.
>  .
>  .
>>> Valgrind is probably what you need.  It doesn't work on windows, but
>>> people use it with wine.
>> Thanks Henry and Sven. I suspect that both FPP and Valgrind will end up
>> being a substantial amount of work for a one-off problem... FPP in
>> particular since it requires Lazarus to build but I'm trying to debug
>> part of the Lazarus build process (bug 0018600).
>>
>> I'm about to start inserting debug messages. I might be some time.
> 
> One possible approximation may be to use the platform specific tools for
> tracing use of the platform APIs (I believe it's strace under Linux?)
> because it may provide some information even before adding the debug
> messages.

Thanks Tomas, had occurred to me but I don't know what's available for 
Windows (and whether it would run on the old version I'm looking at).

If I anticipated doing this regularly I think I'd look at modifying FPP 
to insert "entering" and "exiting" trace code. At a pinch something like 
kdiff3 would probably be able to make sense of a moderate-length output 
trace, but somewhere I've got a better text-comparison algorithm which 
would be more suitable.

-- 
Mark Morgan Lloyd
markMLl .AT. telemetry.co .DOT. uk

[Opinions above are the author's, not those of his employers or colleagues]



More information about the fpc-pascal mailing list