Bug #6209: Marshal.load unable to load previously marshaled data: invalid encoding symbol (EncodingError)

• Status: Closed
• Assignee: -
• Target version: -
• ruby -v: ruby 1.9.3p125 (2012-02-16 revision 34643) [x86_64-darwin11.3.0]

[octplane (Pierre B)]

=begin

= Description

A ruby object (generated with the Classifier gem) is Marshalled on the disk.

When reloaded, the ruby interpreter crashes with the following error:

test-case.rb:8:in load': invalid encoding symbol (EncodingError) from test-case.rb:8:in block in '
from test-case.rb:7:in open' from test-case.rb:7:in '

The marshalled data has been generated using the classifier gem and the provided test-case and test-file

= Expected Behaviour

Ruby reloads the Marshalled data without error

= Step to reproduce

• Install the classifier gem.
• Ensure . does not contain classifier.data
• run (({ruby test-case.rb}))
• Check classifier.data exists
• run (({ruby test-case.rb}))
• the runtime crashed

=end

[nobu (Nobuyoshi Nakada)]

=begin
The minimal code would be:
Marshal.load(Marshal.dump("\xff".force_encoding("ascii-8bit").intern))
=end

[nobu (Nobuyoshi Nakada)]

• Status changed from Open to Closed
• % Done changed from 0 to 100

This issue was solved with changeset r35157.
Pierre, thank you for reporting this issue.
Your contribution to Ruby is greatly appreciated.
May Ruby be with you.

---

• marshal.c (r_symreal): default to ASCII-8BIT for non-ascii symbols,
  otherwise it should be converted to US-ASCII in rb_intern_str() if
  possible. [ruby-core:43762][Bug #6209]

[nobu (Nobuyoshi Nakada)]

Note that fast-stemmer doesn't seem to support 1.9 yet.

I'm having this issue in ruby 1.9.2 p320. Is the bug resolved in any of the available releases?

Thanks in advance!