Rails - Nov 2005 - Strange error(s) on windows only

I have some strange behaviour on Windows XP boxes (the same code on 
Linux doesn''t have the same behaviour). Ruby 1.8.2 and Rails 0.14.2.

I have a single inheritance table nodes with a base class Node 
(acts_as_tree) and some subclasses such as PageNode and FolderNode. If I 
render the tree in an iframe in a set of iframes it doesn''t work, if I 
render it stand alone it works (there are three iframes with different 
url:s for each, nothing strange).

The error is really bizarre. On some of the nodes ruby complains that 
"children is not a method". But it doesn''t complain on all
nodes and
which one differs a lot from request to request. If I do:

rescue
   puts node.class.name
   puts node.kind_of? FolderNode
...

it prints, for example, FolderNode on the first line and then says false 
on the second line (including the times when the first puts actually 
print FolderNode). Very strange.

It feels like there is some problems with script loading. Each node 
subclass has it''s own file but they are all found in the /app/model
dir.
Is this a problem?

Any ideas? Am I vague?

/Marcus

Various different errors gotten when just hitting refresh in the browser:

- undefined method `display_name'' for #<PageNodeType:0x36acc38>
- undefined method `children'' for #<PageNode:0x3761988>
- undefined method `children'' for #<FolderNode:0x37ac5e0>
- uninitialized constant Node  (the line indicates error where 
FolderNode class is defined)
- The single-table inheritance mechanism failed to locate the subclass: 
''PageNodeType''. This error is raised because the column
''type'' is
reserved for storing the class in case of inheritance. Please rename 
this column if you didn''t intend it to be used for storing the 
inheritance class or overwrite NodeType.inheritance_column to use 
another column for that information.
- uninitialized constant FolderNode
- The single-table inheritance mechanism failed to locate the subclass: 
''FolderNodeType''. This error is raised because the column
''type'' is
reserved for storing the class in case of inheritance. Please rename 
this column if you didn''t intend it to be used for storing the 
inheritance class or overwrite NodeType.inheritance_column to use 
another column for that information.

Hi !

2005/11/1, marcus <m-lists@bristav.se>:
> I have some strange behaviour on Windows XP boxes (the same code on
> Linux doesn't have the same behaviour). Ruby 1.8.2 and Rails 0.14.2.
I have ProductCategories in an acts_as_tree.  In my case, it would
complain that has_parent? did not exist.

Don't ask me why it worked, but I had to add that to my class:
class ProductCatalog
  def self.reloadable?
    false
  end
end

Supposedly, that prevents Rails from deleting methods from the class.

Try it, and report here.  I too am on WinXP.  I hadn't tried on Linux,
though.  Might be worth a try.

Bye !
François

_______________________________________________
Rails mailing list
Rails-1W37MKcQCpIf0INCOvqR/iCwEArCW2h5@public.gmane.org
http://lists.rubyonrails.org/mailman/listinfo/rails

Francois Beausoleil wrote:
 > Don''t ask me why it worked, but I had to add that to my class:
 > class ProductCatalog
 >   def self.reloadable?
 >     false
 >   end
 > end
 >

Didn''t help :(

More strangeness:
If I put

sleep(1)

first in the controller method that loads the tree it works!!!! I''ve 
also tried to alter the sleep length and the error comes back when the 
time is smaller.

Why? Can''t webrick handle multiple concurrent requests?

Btw, I don''t consider the problem solved since I guess it can happen 
again anytime, anywhere if I can''t track it down.

/Marcus

marcus wrote:
> Francois Beausoleil wrote:
>  > Don''t ask me why it worked, but I had to add that to my
class:
>  > class ProductCatalog
>  >   def self.reloadable?
>  >     false
>  >   end
>  > end
>  >
> 
> Didn''t help :(
> 
> More strangeness:
> If I put
> 
> sleep(1)
> 
> first in the controller method that loads the tree it works!!!!
I''ve
> also tried to alter the sleep length and the error comes back when the 
> time is smaller.
> 
> Why? Can''t webrick handle multiple concurrent requests?
WEBrick can, but (unless this has changed recently) Rails is not 
thread-safe and uses a mutex to serialise request processing under WEBrick.

regards

   Justin

So this won''t be a problem if running with (F/S)CGI?

/Marcus

Justin Forder wrote:
> marcus wrote:
> 
>> Francois Beausoleil wrote:
>>  > Don''t ask me why it worked, but I had to add that to my
class:
>>  > class ProductCatalog
>>  >   def self.reloadable?
>>  >     false
>>  >   end
>>  > end
>>  >
>>
>> Didn''t help :(
>>
>> More strangeness:
>> If I put
>>
>> sleep(1)
>>
>> first in the controller method that loads the tree it works!!!!
I''ve
>> also tried to alter the sleep length and the error comes back when the 
>> time is smaller.
>>
>> Why? Can''t webrick handle multiple concurrent requests?
> 
> 
> WEBrick can, but (unless this has changed recently) Rails is not 
> thread-safe and uses a mutex to serialise request processing under WEBrick.
> 
> regards
> 
>   Justin
> _______________________________________________
> Rails mailing list
> Rails-1W37MKcQCpIf0INCOvqR/iCwEArCW2h5@public.gmane.org
> http://lists.rubyonrails.org/mailman/listinfo/rails
> 
>

marcus wrote:
> So this won''t be a problem if running with (F/S)CGI?
Marcus, I was just answering your question about WEBrick - and the 
answer would be the same for WEBrick on Linux as for WEBrick on Windows, 
so I don''t know the implications for your original problem. You could 
try InstantRails if you want to get SCGI running quickly on Windows.

regards

    Justin
> 
> /Marcus
> 
> Justin Forder wrote:
> 
>> marcus wrote:
>>
>>> Francois Beausoleil wrote:
>>>  > Don''t ask me why it worked, but I had to add that to
my class:
>>>  > class ProductCatalog
>>>  >   def self.reloadable?
>>>  >     false
>>>  >   end
>>>  > end
>>>  >
>>>
>>> Didn''t help :(
>>>
>>> More strangeness:
>>> If I put
>>>
>>> sleep(1)
>>>
>>> first in the controller method that loads the tree it works!!!!
I''ve
>>> also tried to alter the sleep length and the error comes back when 
>>> the time is smaller.
>>>
>>> Why? Can''t webrick handle multiple concurrent requests?
>>
>>
>>
>> WEBrick can, but (unless this has changed recently) Rails is not 
>> thread-safe and uses a mutex to serialise request processing under 
>> WEBrick.
>>
>> regards
>>
>>   Justin
>> _______________________________________________
>> Rails mailing list
>> Rails-1W37MKcQCpIf0INCOvqR/iCwEArCW2h5@public.gmane.org
>> http://lists.rubyonrails.org/mailman/listinfo/rails
>>
>>
> 
> _______________________________________________
> Rails mailing list
> Rails-1W37MKcQCpIf0INCOvqR/iCwEArCW2h5@public.gmane.org
> http://lists.rubyonrails.org/mailman/listinfo/rails
> 
>