# Where to declare project path for require

**URL:** <https://www.ruby-forum.com/t/where-to-declare-project-path-for-require/164126>\
**Category:** Ruby\
**Created:** [March 28, 2009, 8:35am UTC](https://www.ruby-forum.com/t/where-to-declare-project-path-for-require/164126 "2009-03-28T08:35:35Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![scudco](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/aca169/32.png) [@scudco](https://www.ruby-forum.com/u/scudco)\
**Post date:** [March 28, 2009, 8:35am UTC](https://www.ruby-forum.com/t/where-to-declare-project-path-for-require/164126/1 "2009-03-28T08:35:35Z")

</div>

Untill now whenever i made a script Id put all its files into a  
directory along with any test files. When it came to requiring my own  
files I could just do require “filename” as they were all in the same  
place.

Now ive tried to create a directory structure that a lot of you pros use  
such as

project  
/bin  
/lib  
project.rb  
/project  
…any related files here  
/tests  
test.rb  
/samples  
…any stuff my tests may need e.g. offline webpages to scrape  
etc

though im not planning on releasing any gems i just want a tidier  
structure. however with this comes the headache of require.

At first i simply put this in my project.rb file at the top  
$: \<\< File.expand\_path(File.dirname( **FILE** ) + “/…/lib”)

and that made things ok  
but then when i do tests what do i put in them? do i need to add  
$: \<\< File.expand\_path(File.dirname( **FILE** ) + “/…/lib”)  
again or do i just do an include “/…/lib/project.rb” and let it declare  
the new path??

ideally i want one place where i declare it and then forget all about  
it. I have searched previous posts but none seem to mention what to do  
in your test files. Any one help?

---

<div class="post-metadata">

**Author:** ![scudco](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/aca169/32.png) [@scudco](https://www.ruby-forum.com/u/scudco)\
**Post date:** [March 29, 2009, 7:42am UTC](https://www.ruby-forum.com/t/where-to-declare-project-path-for-require/164126/2 "2009-03-29T07:42:44Z")

</div>

ok ive been hacking away at this trying to figure out why its going  
wrong but cant figure it out.

I understand about the loadpath and that require will go searching for  
your required file in each of the specified directories and stop at the  
first instance it finds.

If i have a three files  
myproject/tests/test.rb

myproject/tests/samples/A.rb  
myproject/tests/samples/B.rb

in test.rb :  
require “samples/A”

in A.rb  
require “B”

that should work shouldnt it??? but in my case it complains saying “no  
such file B.rb”. IT should be able to find it because its in the same  
diretory as A though.

Im going absolutely insane here. can anyone eplain why this is happening  
and what how paths should be set with regards to your main project file  
and test files etc

---

<div class="post-metadata">

**Author:** ![scudco](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/aca169/32.png) [@scudco](https://www.ruby-forum.com/u/scudco)\
**Post date:** [March 29, 2009, 1:52pm UTC](https://www.ruby-forum.com/t/where-to-declare-project-path-for-require/164126/3 "2009-03-29T13:52:12Z")

</div>

Adam A. wrote:

> but in my case it complains saying “no  
> such file B.rb”. IT should be able to find it because its in the same  
> diretory as A though.

No - all files are searched for relative to the $LOAD\_PATH ($:) entries  
only, not the directory which the file doing the requiring is in. Even  
if you have “.” in $LOAD\_PATH (meaning “current directory”), this refers  
to the process’s current working directory, which is often not the same  
as the dir the script is in.

I would do the following:

[myproject/tests/test-helper.rb]

# This sets up all the paths we need

$:.unshift File.expand\_path(File.dirname( **FILE** )+"/…/lib")  
$:.unshift File.expand\_path(File.dirname( **FILE** ))

[myproject/tests/test.rb]  
require File.dirname( **FILE** )+"/test-helper"  
require ‘samples/A’

[myproject/tests/samples/A.rb]  
require ‘samples/B’

The test-helper.rb file sets up absolute paths in $LOAD\_PATH for all the  
directories of interest. So the only thing that a top-level ruby app has  
to do (one which is started directly from the command line) is to  
require test-helper

The example here assumes you have

myproject/lib  
myproject/tests

and adds these two directories to $LOAD\_PATH. So if you do require  
“samples/A” then it will look for /path/to/myproject/tests/samples/A.rb  
and /path/to/myproject/lib/samples/A.rb

Note that samples/A.rb does require ‘samples/B’, because $LOAD\_PATH  
contains /path/to/myproject/tests (and not  
/path/to/myproject/tests/samples)

By using absolute paths in $LOAD\_PATH, it works regardless of which  
directory you are in when you invoke the script. e.g.

# set current working directory to /tmp

cd /tmp  
ruby /path/to/myproject/tests/test.rb

Hope this is clear,

Brian.

---

<div class="post-metadata">

**Author:** ![scudco](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/aca169/32.png) [@scudco](https://www.ruby-forum.com/u/scudco)\
**Post date:** [April 3, 2009, 1:06pm UTC](https://www.ruby-forum.com/t/where-to-declare-project-path-for-require/164126/4 "2009-04-03T13:06:27Z")

</div>

sorry for the late reply, i had a small break from programming.

Just want to thank you very much for your reply. It really has cleared  
it all up for me and ended hours of frustration. Many thanks!

---

<div class="post-metadata">

**Author:** ![scudco](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/aca169/32.png) [@scudco](https://www.ruby-forum.com/u/scudco)\
**Post date:** [March 29, 2009, 2:26pm UTC](https://www.ruby-forum.com/t/where-to-declare-project-path-for-require/164126/5 "2009-03-29T14:26:04Z")

</div>

I don’t know how it’s related to this topic but remember that with  
ruby1.9 we get new require method, which IMHO is very importand:  
require\_relative “file”  
(and as expected it require file relative to **FILE** )  
With ruby1.8 (to be compatible) you can write own impl. of that  
(stolen from  
[http://extensions.rubyforge.org/rdoc/classes/Kernel.html#M000039](http://extensions.rubyforge.org/rdoc/classes/Kernel.html#M000039))  
def require\_relative(path)  
require File.join(File.dirname(caller[0]), path.to\_str)  
end

I think it should be used very often by gem implementators. I don’t  
like if someone modify $: to add some lib internal directories etc.

–  
Pozdrawiam

Rados³aw Bu³at  
[http://radarek.jogger.pl](http://radarek.jogger.pl) - mój blog