rspec users - Jan 2012 - id getting overwritten with 0 when testing

My code that saves a record works fine in development or production, or from the
console. I can take the code in my test and run it in the console, and it works
fine.

But when I run it under a model rspec, the ids are getting set to 0. I?ve traced
it through to where I do:

<Model>.create <params>

where I can see that the id is what I want to set it to in params.

Same problem with rspec 2.6 and 2.8.

Don?t make me switch to Test::Unit. Anyone?

It''d be nice to have a bit of context for this issue.

It''s most likely an issue with your model''s validation... 

Julian

On 25/01/2012, at 5:24 PM, Guyren G Howe wrote:
> My code that saves a record works fine in development or production, or
from the console. I can take the code in my test and run it in the console, and
it works fine.
> 
> But when I run it under a model rspec, the ids are getting set to 0. I?ve
traced it through to where I do:
> 
> <Model>.create <params>
> 
> where I can see that the id is what I want to set it to in params.
> 
> Same problem with rspec 2.6 and 2.8.
> 
> Don?t make me switch to Test::Unit. Anyone?
> _______________________________________________
> rspec-users mailing list
> rspec-users at rubyforge.org
> http://rubyforge.org/mailman/listinfo/rspec-users

On Jan 24, 2012, at 11:27 PM, Julian Leviston wrote:
> On 25/01/2012, at 5:24 PM, Guyren G Howe wrote:
> 
>> My code that saves a record works fine in development or production, or
from the console. I can take the code in my test and run it in the console, and
it works fine.
>> 
>> But when I run it under a model rspec, the ids are getting set to 0.
I?ve traced it through to where I do:
>> 
>> <Model>.create <params>
>> 
>> where I can see that the id is what I want to set it to in params.
>> 
>> Same problem with rspec 2.6 and 2.8.
>> 
>> Don?t make me switch to Test::Unit. Anyone?
> It''d be nice to have a bit of context for this issue.
> 
> It''s most likely an issue with your model''s validation? 
Not sure what else to tell you. I?ve a complex bit of logic I want to exercise
that?s accepting a hierarchy of objects submitted to the application as JSON.
The controller pulls it apart into a hierarchical key-value hash. I?ve a
recursive operation that walks this structure, pulling out individual objects
and saving them.

Everything works fine when I test it manually (e.g. in console). When I run the
same sequence of operations with the same values in console (i.e. I tested it by
copying the values and operations out of the spec into the console), it all
works fine.

But it all fails horribly in rspec because the ids are getting overwritten with
0s. I can get to the point in my code where I hand things over to ActiveRecord,
and the hash I?m giving to create is exactly what I?m after including the id
value.

FWIW, the ids I?m trying to use are UUIDs.

Since I?m entirely sure the hash I?m handing to create is correct, I?m left with
trying to grub around inside ActiveRecord, which I don?t look forward to. So: in
what way does RSpec modify the behavior of ActiveRecord that might bear on this?

On Jan 25, 2012, at 12:43 AM, Guyren G Howe wrote:
> 
> On Jan 24, 2012, at 11:27 PM, Julian Leviston wrote:
> 
>> On 25/01/2012, at 5:24 PM, Guyren G Howe wrote:
>> 
>>> My code that saves a record works fine in development or
production, or from the console. I can take the code in my test and run it in
the console, and it works fine.
>>> 
>>> But when I run it under a model rspec, the ids are getting set to
0. I?ve traced it through to where I do:
>>> 
>>> <Model>.create <params>
>>> 
>>> where I can see that the id is what I want to set it to in params.
>>> 
>>> Same problem with rspec 2.6 and 2.8.
>>> 
>>> Don?t make me switch to Test::Unit. Anyone?
>> It''d be nice to have a bit of context for this issue.
>> 
>> It''s most likely an issue with your model''s
validation?
> 
> Not sure what else to tell you. I?ve a complex bit of logic I want to
exercise that?s accepting a hierarchy of objects submitted to the application as
JSON. The controller pulls it apart into a hierarchical key-value hash. I?ve a
recursive operation that walks this structure, pulling out individual objects
and saving them.
> 
> Everything works fine when I test it manually (e.g. in console). When I run
the same sequence of operations with the same values in console (i.e. I tested
it by copying the values and operations out of the spec into the console), it
all works fine.
> 
> But it all fails horribly in rspec because the ids are getting overwritten
with 0s. I can get to the point in my code where I hand things over to
ActiveRecord, and the hash I?m giving to create is exactly what I?m after
including the id value.
> 
> FWIW, the ids I?m trying to use are UUIDs.
> 
> Since I?m entirely sure the hash I?m handing to create is correct, I?m left
with trying to grub around inside ActiveRecord, which I don?t look forward to.
So: in what way does RSpec modify the behavior of ActiveRecord that might bear
on this?
> _______________________________________________
> rspec-users mailing list
> rspec-users at rubyforge.org
> http://rubyforge.org/mailman/listinfo/rspec-users
http://ariejan.net/2008/08/12/ruby-on-rails-uuid-as-your-activerecord-primary-key

Make sure the column is 16-byte binary

On Wed, 25 Jan 2012, Justin Ko wrote:
> 
>
http://ariejan.net/2008/08/12/ruby-on-rails-uuid-as-your-activerecord-primary-key
> 
> Make sure the column is 16-byte binary
> 
And be sure to update your test database''s schema so that it behaves
the
same as your other environments:

rake db:test:clone_structure

Patrick J. Collins
http://collinatorstudios.com

On Jan 25, 2012, at 1:42 AM, Patrick J. Collins wrote:
> On Wed, 25 Jan 2012, Justin Ko wrote:
> 
>> 
>>
http://ariejan.net/2008/08/12/ruby-on-rails-uuid-as-your-activerecord-primary-key
>> 
>> Make sure the column is 16-byte binary
>> 
> 
> And be sure to update your test database''s schema so that it
behaves the
> same as your other environments:
> 
> rake db:test:clone_structure
You actually want to use "rake db:test:prepare". It does the same
thing as "clone_structure" but doesn''t create the nasty
structure.sql file.

http://stackoverflow.com/questions/7693365/whats-the-difference-between-dbtestclone-dbtestclone-structure-dbtestlo
> 
> Patrick J. Collins
> http://collinatorstudios.com
> _______________________________________________
> rspec-users mailing list
> rspec-users at rubyforge.org
> http://rubyforge.org/mailman/listinfo/rspec-users

See that''s HEAPS better! More information.

Pffft ... not sure what else to tell us!

Okay so NOW to me it sounds a lot like you''re using a non-integer as a
primary key which I wouldn''t do...

And also, could it not be a string coersion issue? (ie params coersion)

Julian

On 25/01/2012, at 6:43 PM, Guyren G Howe wrote:
> 
> On Jan 24, 2012, at 11:27 PM, Julian Leviston wrote:
> 
>> On 25/01/2012, at 5:24 PM, Guyren G Howe wrote:
>> 
>>> My code that saves a record works fine in development or
production, or from the console. I can take the code in my test and run it in
the console, and it works fine.
>>> 
>>> But when I run it under a model rspec, the ids are getting set to
0. I?ve traced it through to where I do:
>>> 
>>> <Model>.create <params>
>>> 
>>> where I can see that the id is what I want to set it to in params.
>>> 
>>> Same problem with rspec 2.6 and 2.8.
>>> 
>>> Don?t make me switch to Test::Unit. Anyone?
>> It''d be nice to have a bit of context for this issue.
>> 
>> It''s most likely an issue with your model''s
validation?
> 
> Not sure what else to tell you. I?ve a complex bit of logic I want to
exercise that?s accepting a hierarchy of objects submitted to the application as
JSON. The controller pulls it apart into a hierarchical key-value hash. I?ve a
recursive operation that walks this structure, pulling out individual objects
and saving them.
> 
> Everything works fine when I test it manually (e.g. in console). When I run
the same sequence of operations with the same values in console (i.e. I tested
it by copying the values and operations out of the spec into the console), it
all works fine.
> 
> But it all fails horribly in rspec because the ids are getting overwritten
with 0s. I can get to the point in my code where I hand things over to
ActiveRecord, and the hash I?m giving to create is exactly what I?m after
including the id value.
> 
> FWIW, the ids I?m trying to use are UUIDs.
> 
> Since I?m entirely sure the hash I?m handing to create is correct, I?m left
with trying to grub around inside ActiveRecord, which I don?t look forward to.
So: in what way does RSpec modify the behavior of ActiveRecord that might bear
on this?
> _______________________________________________
> rspec-users mailing list
> rspec-users at rubyforge.org
> http://rubyforge.org/mailman/listinfo/rspec-users

On Jan 25, 2012, at 3:44 AM, Julian Leviston wrote:
> Okay so NOW to me it sounds a lot like you''re using a non-integer
as a primary key which I wouldn''t do...
I don?t think you?ve tried to write a server app that synchronizes with handheld
apps over an unreliable internet connection. UUIDs make things *so* much easier.

The decision about the pks is made. Am I understanding from folks here that
rspec *won?t work* without integer primary keys? That is a major design flaw, if
true.

On Wed, Jan 25, 2012 at 3:29 PM, Guyren G Howe <gisborne at emailuser.net>
wrote:
> On Jan 25, 2012, at 3:44 AM, Julian Leviston wrote:
>
>> Okay so NOW to me it sounds a lot like you''re using a
non-integer as a primary key which I wouldn''t do...
>
> I don?t think you?ve tried to write a server app that synchronizes with
handheld apps over an unreliable internet connection. UUIDs make things *so*
much easier.
>
> The decision about the pks is made. Am I understanding from folks here that
rspec *won?t work* without integer primary keys? That is a major design flaw, if
true.
RSpec provides a thin wrapper around the testing framework that ships
with Rails and extends T/U. I''m not sure I understand the problem yet,
but I''d be really surprised if it''s anything RSpec is doing or
failing
to do.

What versions of RSpec and Rails are you using?

On Wed, Jan 25, 2012 at 3:37 PM, David Chelimsky <dchelimsky at gmail.com>
wrote:
> On Wed, Jan 25, 2012 at 3:29 PM, Guyren G Howe <gisborne at
emailuser.net> wrote:
>> On Jan 25, 2012, at 3:44 AM, Julian Leviston wrote:
>>
>>> Okay so NOW to me it sounds a lot like you''re using a
non-integer as a primary key which I wouldn''t do...
>>
>> I don?t think you?ve tried to write a server app that synchronizes with
handheld apps over an unreliable internet connection. UUIDs make things *so*
much easier.
>>
>> The decision about the pks is made. Am I understanding from folks here
that rspec *won?t work* without integer primary keys? That is a major design
flaw, if true.
>
> RSpec provides a thin wrapper around the testing framework that ships
> with Rails and extends T/U. I''m not sure I understand the problem
yet,
> but I''d be really surprised if it''s anything RSpec is
doing or failing
> to do.
>
> What versions of RSpec and Rails are you using?
Just looked back at your initial email and see you cited rspec-2.8,
but the way Rails handles incoming params in tests changed in either
3.1 or 3.2 (I have to check). Which rails version specifically?

On Jan 25, 2012, at 1:45 PM, David Chelimsky wrote:
> Just looked back at your initial email and see you cited rspec-2.8,
> but the way Rails handles incoming params in tests changed in either
> 3.1 or 3.2 (I have to check). Which rails version specifically?
3.1.

Since I was *always* going to be providing a UUID for the pk, I didn?t even
bother changing the default value for the pk in the schema. I?m going to try
fixing that and see if it?s any better.

On Wed, Jan 25, 2012 at 3:49 PM, Guyren G Howe <gisborne at emailuser.net>
wrote:
> On Jan 25, 2012, at 1:45 PM, David Chelimsky wrote:
>
>> Just looked back at your initial email and see you cited rspec-2.8,
>> but the way Rails handles incoming params in tests changed in either
>> 3.1 or 3.2 (I have to check). Which rails version specifically?
>
> 3.1.
>
> Since I was *always* going to be providing a UUID for the pk, I didn?t even
bother changing the default value for the pk in the schema. I?m going to try
fixing that and see if it?s any better.
I''d start by debugging to see where the wheels fall off. Are you
familiar/comfortable with Ruby''s debugger?

On Jan 25, 2012, at 1:56 PM, David Chelimsky wrote:
> I''d start by debugging to see where the wheels fall off. Are you
> familiar/comfortable with Ruby''s debugger?
Sure.

What?s happening is that during the save process, I get to field_changed? in
dirty.rb, which does

	value = column.type_cast(value)

when I look at column here, it believes @sql_type = ?integer?. Which seems
weird.

So at this point, it occurs to me to check this against my test database, and
the columns in the test database are indeed integers! How does this come to be?

On Jan 25, 2012, at 4:50 PM, Guyren G Howe wrote:
> On Jan 25, 2012, at 1:56 PM, David Chelimsky wrote:
> 
>> I''d start by debugging to see where the wheels fall off. Are
you
>> familiar/comfortable with Ruby''s debugger?
> 
> Sure.
> 
> What?s happening is that during the save process, I get to field_changed?
in dirty.rb, which does
> 
> 	value = column.type_cast(value)
> 
> when I look at column here, it believes @sql_type = ?integer?. Which seems
weird.
> 
> So at this point, it occurs to me to check this against my test database,
and the columns in the test database are indeed integers! How does this come to
be?
And note that

	rake db:test:load

doesn?t change anything.

On Jan 25, 2012, at 6:20 PM, Guyren Howe wrote:
> On Jan 25, 2012, at 4:50 PM, Guyren G Howe wrote:
> 
>> On Jan 25, 2012, at 1:56 PM, David Chelimsky wrote:
>> 
>>> I''d start by debugging to see where the wheels fall off.
Are you
>>> familiar/comfortable with Ruby''s debugger?
>> 
>> Sure.
>> 
>> What?s happening is that during the save process, I get to
field_changed? in dirty.rb, which does
>> 
>> 	value = column.type_cast(value)
>> 
>> when I look at column here, it believes @sql_type = ?integer?. Which
seems weird.
>> 
>> So at this point, it occurs to me to check this against my test
database, and the columns in the test database are indeed integers! How does
this come to be?
> 
> And note that
> 
> 	rake db:test:load
> 
> doesn?t change anything.
> _______________________________________________
> rspec-users mailing list
> rspec-users at rubyforge.org
> http://rubyforge.org/mailman/listinfo/rspec-users
Please paste the table from your schema.rb file.

On Wed, Jan 25, 2012 at 7:20 PM, Guyren Howe <guyren at mac.com>
wrote:
> On Jan 25, 2012, at 4:50 PM, Guyren G Howe wrote:
>
>> On Jan 25, 2012, at 1:56 PM, David Chelimsky wrote:
>>
>>> I''d start by debugging to see where the wheels fall off.
Are you
>>> familiar/comfortable with Ruby''s debugger?
>>
>> Sure.
>>
>> What?s happening is that during the save process, I get to
field_changed? in dirty.rb, which does
>>
>> ? ? ? value = column.type_cast(value)
>>
>> when I look at column here, it believes @sql_type = ?integer?. Which
seems weird.
>>
>> So at this point, it occurs to me to check this against my test
database, and the columns in the test database are indeed integers! How does
this come to be?
>
> And note that
>
> ? ? ? ?rake db:test:load
>
> doesn?t change anything.
I''m not sure what rake db:test:load actually does, but `RAILS_ENV=test
rake db:reset` should probably do the trick.

What db are you using, btw?