Hi, I was wondering if anybody had any advice on the following subjects:

1. How do you get "print" statements executed by the server? For instance,
I'm running Sybase, and on 

   dump database foo to $file

the server outputs a series of print statements as it does the backup. I
want to intercept these statements, but so far I can seem to see how to do
so.

2. The log mechanism used by $dbh->trace() bypasses all of perls normal file
handling mechanisms. It doesnt print via *STDOUT/*STDERR but via perls
_internal_ equivelent. This means that any output from trace cannot be
intercepted by tieing these filehandles, as works almost everywhere else.
Nor can you specify '&1' or '&2' or '-' to get it to do so (you simply end
up with files with these as the name), nor can you provide a filehandle to
print to.  This is a problem for me because this doesnt play very nicely
with other logging or a modular approach to tracing/logging. 

On a feature level I was wondering if there was any chance of being able to
provide a filehandle instead of a filename for trace() to use, or some other
mechanism by which trace output can be intercepted. Perhaps if a subref was
provided to trace() instead of a filename it would call the subref with the
current trace output? This way it could be handled as the end user chooses? 

On a workaround level, can anybody suggest something?

Cheers,
Yves
ps: Tim, thanks for all your effort. A very powerful tool indeed.






Reply via email to