Length of backtrace displayed by ruby

[Sylvain Joyeux]

Is there a way to remove the maximum backtrace length displayed by the
interpreter when an exception terminates the process ? It makes the backtrace
totally useless in situations where it is long *and* the interesting calls
are in the middle.

Thanks

Sylvain

 

[Yukihiro Matsumoto]

Hi,

In message "Re: Length of backtrace displayed by ruby"

|Is there a way to remove the maximum backtrace length displayed by the
|interpreter when an exception terminates the process ? It makes the backtrace
|totally useless in situations where it is long *and* the interesting calls
|are in the middle.

Not yet. Do you (or anybody else) have any good API suggestion?

matz.

 

[ara.t.howard]

Hi,

In message "Re: Length of backtrace displayed by ruby"

|Is there a way to remove the maximum backtrace length displayed by the
|interpreter when an exception terminates the process ? It makes the backtrace
|totally useless in situations where it is long *and* the interesting calls
|are in the middle.

Not yet. Do you (or anybody else) have any good API suggestion?

matz.

Exception.backtrace_display_limit = 42

??

-a

 

[Jim Weirich]

Is there a way to remove the maximum backtrace length displayed by the
interpreter when an exception terminates the process ? It makes the
backtrace
totally useless in situations where it is long *and* the interesting
calls
are in the middle.

IIRC, this only happens when the exception is handled by the top level
interpreter. If you catch the exception yourself at the top level you
can print out whatever you want, including the full stack trace.

 

[Stephen Kellett]

Exception.backtrace_display_limit = 42

Exception.setFilter(regularExpression)

where the matching items are removed from the backtrace.

Stephen

 

[ara.t.howard]

Exception.setFilter(regularExpression)

where the matching items are removed from the backtrace.

what would be matched?

e.to_s, e.message.to_s, or e.backtrace.any?{|b|}

regards.

-a

 

[Sylvain Joyeux]

IIRC, this only happens when the exception is handled by the top level

interpreter. If you catch the exception yourself at the top level you
can print out whatever you want, including the full stack trace.

Yes, that's what I did as a workaround. But you have to do it for every
thread and every fork (which in my case is 4 different places)

 

[Stephen Kellett]

what would be matched?

e.to_s, e.message.to_s, or e.backtrace.any?{|b|}

Class and method name was what I was thinking of.