Customing the Exception class

ruby-talk
1

Hi all,

I have a program where I want to send all errors to a file. So, I did this:

class Exception
    alias :old_init :initialize
    def initialize(arg)
       old_init(arg)
       File.open("log.txt","a+"){ |fh| fh.puts("Error: #{arg}") }
    end
end

This works for most errors, but it doesn't seem to work when the error is a SystemCallError.

For example, this writes to the log as expected:

print "Some crazy division: "
p 1/0

But this does not:

File.open("non-existant file")

The latter raises an Errno::ENOENT.

Why, if SystemCallError is a subclass of Exception, does my approach not work? How do I make this work as expected?

Regards,

Dan

2

maybe new?

   harp:~ > cat a.rb
   class Exception
     class << self
       alias ____new new
       def new(*a, &b)
         ret = ____new(*a, &b)
         STDERR.puts "new: <#{ a.inspect }>"
         ret
       end
     end
     alias ____init initialize
     def initialize(*a, &b)
       ret = ____init(*a, &b)
       STDERR.puts "initialize: <#{ a.inspect }>"
       ret
     end
   end

   SystemCallError::new 42

   harp:~ > ruby a.rb
   new: <[42]>

not sure why...

-a

···

On Sat, 14 May 2005, Daniel Berger wrote:

Hi all,

I have a program where I want to send all errors to a file. So, I did this:

class Exception
   alias :old_init :initialize
   def initialize(arg)
      old_init(arg)
      File.open("log.txt","a+"){ |fh| fh.puts("Error: #{arg}") }
   end
end

This works for most errors, but it doesn't seem to work when the error
is a SystemCallError.

For example, this writes to the log as expected:

print "Some crazy division: "
p 1/0

But this does not:

File.open("non-existant file")

The latter raises an Errno::ENOENT.

Why, if SystemCallError is a subclass of Exception, does my approach not
work? How do I make this work as expected?

--

email :: ara [dot] t [dot] howard [at] noaa [dot] gov
phone :: 303.497.6469
renunciation is not getting rid of the things of this world, but accepting
that they pass away. --aitken roshi

===============================================================================

3

Hi all,

I have a program where I want to send all errors to a file. So, I did this:

[snip]

This works for most errors, but it doesn't seem to work when the error
is a SystemCallError.

For example, this writes to the log as expected:

print "Some crazy division: "
p 1/0

But this does not:

File.open("non-existant file")

The latter raises an Errno::ENOENT.

Why, if SystemCallError is a subclass of Exception, does my approach not
work? How do I make this work as expected?

maybe new?

[snip]

not sure why...

SystemCallError does not use Exception's initialize:

     rb_eSystemCallError = rb_define_class("SystemCallError", rb_eStandardError);
     rb_define_method(rb_eSystemCallError, "initialize", syserr_initialize, -1);

···

On 13 May 2005, at 14:35, Ara.T.Howard@noaa.gov wrote:

On Sat, 14 May 2005, Daniel Berger wrote:

--
Eric Hodel - drbrain@segment7.net - http://segment7.net
FEC2 57F1 D465 EB15 5D6E 7C11 332A 551C 796C 9F04

4

Eric Hodel wrote:

···

On 13 May 2005, at 14:35, Ara.T.Howard@noaa.gov wrote:

> On Sat, 14 May 2005, Daniel Berger wrote:
>
>
>> Hi all,
>>
>> I have a program where I want to send all errors to a file. So, I

>> did this:

[snip]

>> This works for most errors, but it doesn't seem to work when the
>> error
>> is a SystemCallError.
>>
>> For example, this writes to the log as expected:
>>
>> print "Some crazy division: "
>> p 1/0
>>
>> But this does not:
>>
>> File.open("non-existant file")
>>
>> The latter raises an Errno::ENOENT.
>>
>> Why, if SystemCallError is a subclass of Exception, does my
>> approach not
>> work? How do I make this work as expected?
>
> maybe new?

[snip]

> not sure why...

SystemCallError does not use Exception's initialize:

     rb_eSystemCallError = rb_define_class("SystemCallError",
rb_eStandardError);
     rb_define_method(rb_eSystemCallError, "initialize",
syserr_initialize, -1);

Yuck. Could it be altered to call super? Or would that cause some
hideous side effect?

Regards

Dan

PS - Sorry if this is a double post - google groups went goofy on me
for a minute.

5

Hi,

At Sat, 14 May 2005 10:30:29 +0900,
Daniel Berger wrote in [ruby-talk:142610]:

> SystemCallError does not use Exception's initialize:
>
> rb_eSystemCallError = rb_define_class("SystemCallError",
> rb_eStandardError);
> rb_define_method(rb_eSystemCallError, "initialize",
> syserr_initialize, -1);

Accurately, syserr_initialize() directly calls exc_initialize()
which implements Execption#initialize.

Yuck. Could it be altered to call super? Or would that cause some
hideous side effect?

Just for efficiency, I guess.

Index: error.c

···

===================================================================
RCS file: /cvs/ruby/src/ruby/error.c,v
retrieving revision 1.107
diff -U2 -p -r1.107 error.c
--- error.c 18 Mar 2005 03:17:27 -0000 1.107
+++ error.c 14 May 2005 02:32:25 -0000
@@ -387,5 +387,5 @@ exc_exception(argc, argv, self)
     if (argc == 1 && self == argv[0]) return self;
     exc = rb_obj_clone(self);
- exc_initialize(argc, argv, exc);
+ rb_obj_call_init(exc, argc, argv);

     return exc;
@@ -581,5 +581,5 @@ exit_initialize(argc, argv, exc)
   --argc;
     }
- exc_initialize(argc, argv, exc);
+ rb_call_super(argc, argv);
     rb_iv_set(exc, "status", status);
     return exc;
@@ -661,5 +661,5 @@ name_err_initialize(argc, argv, self)

     name = (argc > 1) ? argv[--argc] : Qnil;
- exc_initialize(argc, argv, self);
+ rb_call_super(argc, argv);
     rb_iv_set(self, "name", name);
     return self;
@@ -966,5 +966,5 @@ syserr_initialize(argc, argv, self)
   mesg = rb_str_new2(err);
     }
- exc_initialize(1, &mesg, self);
+ rb_call_super(1, &mesg);
     rb_iv_set(self, "errno", error);
     return self;

--
Nobu Nakada

6

Will this patch be applied to the 1.8 branch? I'd like to see it in
1.8.3 but I understand if you'd rather merge it into 1.9 instead.

Regards,

Dan

···

nobu.nok...@softhome.net wrote:

Hi,

At Sat, 14 May 2005 10:30:29 +0900,
Daniel Berger wrote in [ruby-talk:142610]:
> > SystemCallError does not use Exception's initialize:
> >
> > rb_eSystemCallError = rb_define_class("SystemCallError",
> > rb_eStandardError);
> > rb_define_method(rb_eSystemCallError, "initialize",
> > syserr_initialize, -1);

Accurately, syserr_initialize() directly calls exc_initialize()
which implements Execption#initialize.

> Yuck. Could it be altered to call super? Or would that cause some
> hideous side effect?

Just for efficiency, I guess.

Index: error.c

RCS file: /cvs/ruby/src/ruby/error.c,v
retrieving revision 1.107
diff -U2 -p -r1.107 error.c
--- error.c 18 Mar 2005 03:17:27 -0000 1.107
+++ error.c 14 May 2005 02:32:25 -0000
@@ -387,5 +387,5 @@ exc_exception(argc, argv, self)
     if (argc == 1 && self == argv[0]) return self;
     exc = rb_obj_clone(self);
- exc_initialize(argc, argv, exc);
+ rb_obj_call_init(exc, argc, argv);

     return exc;
@@ -581,5 +581,5 @@ exit_initialize(argc, argv, exc)
   --argc;
     }
- exc_initialize(argc, argv, exc);
+ rb_call_super(argc, argv);
     rb_iv_set(exc, "status", status);
     return exc;
@@ -661,5 +661,5 @@ name_err_initialize(argc, argv, self)

     name = (argc > 1) ? argv[--argc] : Qnil;
- exc_initialize(argc, argv, self);
+ rb_call_super(argc, argv);
     rb_iv_set(self, "name", name);
     return self;
@@ -966,5 +966,5 @@ syserr_initialize(argc, argv, self)
   mesg = rb_str_new2(err);
     }
- exc_initialize(1, &mesg, self);
+ rb_call_super(1, &mesg);
     rb_iv_set(self, "errno", error);
     return self;

--
Nobu Nakada

7

Hi,

···

In message "Re: Customing the Exception class" on Sat, 14 May 2005 11:35:20 +0900, nobu.nokada@softhome.net writes:

Yuck. Could it be altered to call super? Or would that cause some
hideous side effect?

Just for efficiency, I guess.

Hmm, OK, commit the patch, please.

              matz.