# Auto-generated find queries adding bogus WHERE clauses

**URL:** <https://www.ruby-forum.com/t/auto-generated-find-queries-adding-bogus-where-clauses/53570>\
**Category:** Rails\
**Created:** [February 10, 2006, 3:34am UTC](https://www.ruby-forum.com/t/auto-generated-find-queries-adding-bogus-where-clauses/53570 "2006-02-10T03:34:17Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![mjgp2](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/eb8c5e/32.png) [@mjgp2](https://www.ruby-forum.com/u/mjgp2)\
**Post date:** [February 10, 2006, 3:34am UTC](https://www.ruby-forum.com/t/auto-generated-find-queries-adding-bogus-where-clauses/53570/1 "2006-02-10T03:34:17Z")

</div>

[Sorry for the poor subject; this is really hard to explain]

I’m trying to run a search over several objects which themselves contain  
multiple model objects, like so:

class Collection \<\< ActiveRecord::Base  
def findpkg  
self.repositories.each do |r|  
p = r.packages.find(:all, :conditions =\> “name like ‘%pkg%’”)  
end  
end  
end

Unfortunately, AR proceeds to generate SQL that looks like this:

Package Load (0.000548) SELECT \* FROM packages WHERE  
(packages.repository\_id = 3 AND (name like ‘%pkg%’))  
Package Load (0.000167) SELECT \* FROM packages WHERE  
(packages.repository\_id = 4 AND (packages.repository\_id = 3  
AND (name like ‘%pkg%’)))  
Package Load (0.000151) SELECT \* FROM packages WHERE  
(packages.repository\_id = 5 AND (packages.repository\_id = 4  
AND (packages.repository\_id = 3 AND (name like ‘%pkg%’))))  
Package Load (0.000160) SELECT \* FROM packages WHERE  
(packages.repository\_id = 13 AND (packages.repository\_id = 5  
AND (packages.repository\_id = 4 AND (packages.repository\_id = 3  
AND (name like ‘%pkg%’)))))

Note that each successive query prepends an extra clause, instead of  
replacing the previous repository\_id clause.

Has anyone seen anything like this? I can provide relevant chunks of  
code  
on request, but I’m not doing anything exciting in my models – just  
standard sorts of has\_many relationships. I’ll turn this into a bug  
report  
if needed, I just want to get a reality check before I go nuts.

- Matt

---

<div class="post-metadata">

**Author:** ![mjgp2](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/eb8c5e/32.png) [@mjgp2](https://www.ruby-forum.com/u/mjgp2)\
**Post date:** [February 11, 2006, 6:58am UTC](https://www.ruby-forum.com/t/auto-generated-find-queries-adding-bogus-where-clauses/53570/2 "2006-02-11T06:58:07Z")

</div>

[Please respect MFT to rails-core; I think most of the discussion of  
this,  
if any, will be more topical there]

On Fri, Feb 10, 2006 at 01:33:32PM +1100, Matthew P. wrote:

> class Collection \<\< ActiveRecord::Base  
> def findpkg  
> self.repositories.each do |r|  
> p = r.packages.find(:all, :conditions =\> “name like ‘%pkg%’”)  
> end  
> end  
> end

The bug is that HasManyAssociation#find modifies it’s arguments, so the  
above (slightly fictionalised) variant is misleading; the real code is:

def findpkg(\*args)  
self.repositories.each do |r|  
p = r.packages.find(\*args)  
# Do things with p  
end  
end

Adding “puts args.inspect” before and after the call to find() shows  
that  
args is in fact being modified, by the fact that the options object is  
the  
same as args[1] on the entry to HasManyAssociation#find. This simple  
patch  
makes the whole problem go away:

------8\<------  
— has\_many\_association.rb 2006-02-10 16:37:32.000000000 +1100  
+++ has\_many\_association.rb.orig 2006-02-10 16:37:21.000000000  
+1100  
@@ -51,7 +51,7 @@  
end

```
   def find(*args)

```

- 

```
   options = Base.send(:extract_options_from_args!, args).dup

```

- 

```
   options = Base.send(:extract_options_from_args!, args)

   # If using a custom finder_sql, scan the entire collection.
   if @options[:finder_sql]

```

------\>8------

I’m planning on reporting this in Trac, but I’m curious about people’s  
opinions – is it reasonable to assume that arguments which appear, on  
the  
surface, to be read-only may be modified by called methods?

- Matt