JRuby

mkdir_p wrongfully encodes directory names when ran from a Jar

Details

  • Type: Bug Bug
  • Status: Resolved Resolved
  • Priority: Blocker Blocker
  • Resolution: Fixed
  • Affects Version/s: JRuby 1.6RC1
  • Fix Version/s: JRuby 1.7.0.pre1
  • Component/s: Java Integration
  • Labels:
    None
  • Environment:
  • Testcase included:
    yes
  • Number of attachments :
    0

Description

This bug blocks the tests I would like to implement for JRUBY-5345 (we might need to reopen once we have a reliable test suite for this case).

When creating a directory with an "hié" (UTF-8 e-accute-accent) from within a Jar, will instead create a directory called "hi%89" (URL Encoded version of MacRoman e-accute-accent).
To prevent any source-file encoding issue, I manually pack the "é" accent from its binary representation with [0xc3, 0xa9].pack('C*') and I freeze the string.

In https://github.com/lucasdicioccio/jruby/commits/accents

I've modified test_jar_complete.rb:

  • added a test case for jar run from paths with accents (JRUBY-5345)
  • adding an assertion on the content of the test directory content (this bug report)

I've also added a similar test in test_jruby_dirnames.rb . It executes without failure/errors with ./bin/jruby from the same build. Thus I think the issue is JAR-related.

Activity

Ascending order - Click to sort in descending order
Hide
Permalink
Hiro Asari added a comment -

Hi, Lucas,

I apologize for not reviewing this sooner.

It sounds like you had a patch, but it's been deleted. Do you still have it?

Show
Hiro Asari added a comment - Hi, Lucas, I apologize for not reviewing this sooner. It sounds like you had a patch, but it's been deleted. Do you still have it?
Hide
Permalink
lucas dicioccio added a comment -

Hi,
Since the bug has been fixed, I think I'm affraid I've deleted the branch. Now that I think about it, along with my temptatives to fix the bug, I had a test case that was worth saving (the only thing I did was adding a directory with a weird file name to the jar test in test/test_jar_complete.rb : I built the string for the path with accent using ['c3','a9'].pack('H2H2') (for 'é':utf8)).
I'll look if I have a backup in one of my test VMs, but I guess I won't.

Show
lucas dicioccio added a comment - Hi, Since the bug has been fixed, I think I'm affraid I've deleted the branch. Now that I think about it, along with my temptatives to fix the bug, I had a test case that was worth saving (the only thing I did was adding a directory with a weird file name to the jar test in test/test_jar_complete.rb : I built the string for the path with accent using ['c3','a9'] .pack('H2H2') (for 'é':utf8)). I'll look if I have a backup in one of my test VMs, but I guess I won't.
Hide
Permalink
Charles Oliver Nutter added a comment -

We'll call this fixed. It's very likely that we picked up some Dir tests that deal with unicode since then, and even if not...this isn't a blocker and isn't broken anymore.

Show
Charles Oliver Nutter added a comment - We'll call this fixed. It's very likely that we picked up some Dir tests that deal with unicode since then, and even if not...this isn't a blocker and isn't broken anymore.

People

  • Assignee:
    Charles Oliver Nutter
    Reporter:
    lucas dicioccio
Vote (0)
Watch (1)

Dates

  • Created:
    Updated:
    Resolved: