8 Replies Latest reply on Jun 13, 2007 11:55 AM by julien1

URLNavigator Tests

mwringe May 8, 2007 1:37 PM

I have been updating the urlnavigator testcases and fixing some issues with the file and jar urlnavigators.

Would it make sense to test situations such as:
1) invalid jars: jars that have elements added or removed from them
2) files with symbolic links: possible situation in which a loop could occur

The jar urlnavigator still has some issues with it, its a bit of a pain the way jars are handled in java.

1. Re: URLNavigator Tests

julien1 May 11, 2007 10:02 AM (in response to mwringe)

for point 1/ : yes it is a must to have

normally the implementation should deal with it as it simulates missing stuff using a stack

for point 2/ : what would be the behavior ? follow the link or skip it ?
Actions
2. Re: URLNavigator Tests

mwringe May 11, 2007 10:16 AM (in response to mwringe)

"julien@jboss.com" wrote:

for point 2/ : what would be the behavior ? follow the link or skip it ?

If we follow the links we could get into an infinite loop situation. For example
/a1/
/a1/b1 -> /a1/ [b1 links back to a1]

The other issue is how can this test be integrated into the testsuite? I don't believe Windows has the concept of links (although I could be wrong)

I also think I need to look a bit more into how java handles symlinks.
Actions
3. Re: URLNavigator Tests

julien1 May 11, 2007 10:25 AM (in response to mwringe)

Ok, probably that it is hard to unit test because it depends on the target environment.
Actions
4. Re: URLNavigator Tests

prabhat.jha May 11, 2007 11:02 AM (in response to mwringe)

Cant we make this symlink test run only when OS is non-windows? I found this http://shell-shocked.org/article.php?id=284 for symlink in windows but it does not seem cool.
Actions
5. Re: URLNavigator Tests

mwringe Jun 12, 2007 9:41 PM (in response to mwringe)

Sorry, I have actually had this done for a while and should have spoke up before now.

I made a couple of changes to the code outside of the testcase. I implemented the option to filter on file name (he option was there before but it just didn't do anything).
I also added the option to visit the root of a jar now. In order to do this, I had to add a fake jar entry, '/', since this doesn't actually exist in the jar. This causes the current jar entry test to fail. Since this entry doesn't actually exist, is it ok to add it?
Actions
6. Re: URLNavigator Tests

julien1 Jun 13, 2007 8:52 AM (in response to mwringe)

"mwringe" wrote:
I implemented the option to filter on file name (he option was there before but it just didn't do anything).

Good.

"mwringe" wrote:
I also added the option to visit the root of a jar now. In order to do this, I had to add a fake jar entry, '/', since this doesn't actually exist in the jar. This causes the current jar entry test to fail. Since this entry doesn't actually exist, is it ok to add it?

I don't understand, can you explain more precisely ?
Actions
7. Re: URLNavigator Tests

mwringe Jun 13, 2007 11:00 AM (in response to mwringe)

The root of a jar, '/', doesn't actually exist as an entry in the jar. I added a "/" atom object in the jarentryinfo class. So now when you go to navigate the jar using the jarentryinfo class, the root element appears to exist. This allows for visiting a jar starting at the root.

The jar entry test would have to be updated since it doesn't expect the root element to be there.
Actions
8. Re: URLNavigator Tests

julien1 Jun 13, 2007 11:55 AM (in response to mwringe)

the sequence of generated events that is expected by the tests has to take in account now the presence of "/" entry.

it looks normal to me.

I think you can go ahead and commit your changes
Actions

Go to original post