Project

General

Profile

Actions

Bug #2647

closed

Lack of testing for String#split

Added by hgs (Hugh Sasse) almost 15 years ago. Updated over 13 years ago.

Status:
Closed
Assignee:
-
Target version:
-
ruby -v:
ruby 1.8.7 (2008-08-11 patchlevel 72) [i386-cygwin]
[ruby-core:27794]

Description

=begin
Having raised the issue of docs for String#split, I see there seem to be no tests. The attached patch seems to fix this [i.e is tested against the cygwin build] and is based on the documentation (as amended by my recent suggestion). Hope this is OK. It is a patch against the svn ruby_1_8.

This isn't exactly a feature for Ruby, it's not drawn from Ruby Spec, and it isn't a backport, so that only leaves "bug". I'm not sure it's that, it's more an "enhancement". Unless lack of test coverage is a bug.
=end


Files

test_string_patch (2.74 KB) test_string_patch Patch to add tests for String#split hgs (Hugh Sasse), 01/25/2010 09:09 PM
Actions #1

Updated by vvs (Vladimir Sizikov) almost 15 years ago

=begin
Actually, it would be really nice to put such common/core tests into RubySpec, so that all various Ruby implementations would benefit from it. And since MRI 1.9.2 goal to pass 100% RubySpec as well, that would benefit MRI as well.

In fact, there are quite some tests for String#split in RubySpec already:
http://github.com/rubyspec/rubyspec/blob/master/core/string/split_spec.rb

If some of your tests provide better coverage in some cases, let me know and we could get them incorporated in RubySpec.

For more details about RubySpec, take a look here: http://rubyspec.org/
Main repository: http://github.com/rubyspec/rubyspec
=end

Actions #2

Updated by headius (Charles Nutter) almost 15 years ago

=begin
Here here! The tests in RubySpec are vastly better than those in MRI,
and I strongly suggest the Ruby core team should be running RubySpec
to test MRI (perhaps in addition to existing tests, since not all
libraries are well covered by RubySpec). That's the bottom line.

As it stands now, MRI seems to be the only implementation not using
RubySpec on a daily basis. Doesn't that seem a little odd?

On Mon, Jan 25, 2010 at 2:05 PM, Vladimir Sizikov wrote:

Issue #2647 has been updated by Vladimir Sizikov.

Actually, it would be really nice to put such common/core tests into RubySpec, so that all various Ruby implementations would benefit from it. And since MRI 1.9.2 goal to pass 100% RubySpec as well, that would benefit MRI as well.

In fact, there are quite some tests for String#split in RubySpec already:
http://github.com/rubyspec/rubyspec/blob/master/core/string/split_spec.rb

If some of your tests provide better coverage in some cases, let me know and we could get them incorporated in RubySpec.

For more details about RubySpec, take a look here: http://rubyspec.org/
Main repository: http://github.com/rubyspec/rubyspec

http://redmine.ruby-lang.org/issues/show/2647


http://redmine.ruby-lang.org

=end

Actions #3

Updated by naruse (Yui NARUSE) almost 15 years ago

=begin
RubySpec is suitable for Implementation independent tests.

Ruby 1.9 already has a test for String#split but 1.8 doesn't.

As it stands now, MRI seems to be the only implementation not using
RubySpec on a daily basis. Doesn't that seem a little odd?

Current RubySpec for Ruby 1.9 has too many failure because of tests.
(yeah, we know we should fix tests; that's why we postponed 1.9.2 release)

Another reason is test-all is white box tests for MRI.
=end

Actions #4

Updated by naruse (Yui NARUSE) almost 15 years ago

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

=begin
This issue was solved with changeset r26410.
Hugh, thank you for reporting this issue.
Your contribution to Ruby is greatly appreciated.
May Ruby be with you.

=end

Actions #5

Updated by hgs (Hugh Sasse) almost 15 years ago

=begin
OK, can someone please document the proper procedure for this, then? At present

http://www.ruby-lang.org/en/community/ruby-core/#patching-ruby

says we can send patches to the list, or put them in redmine, but I'm getting the impression some people say they get lost unless they are in redmine. Posting them to redmine means the attached patch doesn't show up as an attachement on the list. Redmine refers to RubySpec, and you're telling me that I should put tests into there. But the Documentation for Ruby is based on Rdoc, and my patch agrees with that. Yet, as my recent post about Zlib::GzipReader points out, we also have docs that differ from the Rdoc in RD format, still being shipped out with Ruby.

Can we have a process that is DRY, please?

Vladimir Sizikov wrote:
"If some of your tests provide better coverage in some cases, let me know"

The link is there in the body of the redmine posting. If you are asking me to read up on RubySpec, please could you look at the 93 line patch? It is mostly asserts to try and avoid dependency on Array#each, an entirely different fundamental type. So as code goes it is light reading, if rather dull :-).
=end

Actions #6

Updated by hgs (Hugh Sasse) almost 15 years ago

=begin
On Mon, 25 Jan 2010, Yui NARUSE wrote:

Issue #2647 has been updated by Yui NARUSE.

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

This issue was solved with changeset r26410.
Hugh, thank you for reporting this issue.
Your contribution to Ruby is greatly appreciated.
May Ruby be with you.

Thank you, but when I did svn update my patch seems to have
been added three times, redefining the method test_split
twice as a result. I can't see how to re-open this on
redmine to flag that up. Do I submit a new bug or what?

     Hugh

=end

Actions #7

Updated by naruse (Yui NARUSE) almost 15 years ago

=begin
Sorry, I missed the patch; thank you
=end

Actions #8

Updated by hgs (Hugh Sasse) almost 15 years ago

=begin
On Tue, 26 Jan 2010, Yui NARUSE wrote:

Issue #2647 has been updated by Yui NARUSE.

Sorry, I missed the patch; thank you

No problem, that looks right to me now. Thank you for your trouble.

Had a look at Ruby Spec: Seems to require a number of things not
shipped with Ruby, rspec[?] and a Mocking Framework, requires a
github account to be able to contribute. The tests in RubySpec
don't seem to include Regexps with captures, which mine do. I
don't yet know how to rewrite mine to fit Ruby Spec syntax.

     Hugh

http://redmine.ruby-lang.org/issues/show/2647


http://redmine.ruby-lang.org

=end

Actions #9

Updated by mame (Yusuke Endoh) almost 15 years ago

=begin
Hi,

2010/1/25 Charles Oliver Nutter :

Here here! The tests in RubySpec are vastly better than those in MRI,

Indeed, such common tests are also suitable for RubySpec and tests in
MRI 1.8 are really very short, but I think that current tests in 1.9
is not so bad. It achieves decent coverage.

http://dame.dyndns.org:7001/20100125/ruby/
http://dame.dyndns.org/misc/misc/test-based-code-reading.ppt (at RubyKaigi 2008)

As it stands now, MRI seems to be the only implementation not using
RubySpec on a daily basis. Doesn't that seem a little odd?

Yugui said that 1.9.2 would pass rubyspec (though I guess it is the
showstopper of 1.9.2 release.)

--
Yusuke ENDOH

=end

Actions #10

Updated by rue (Eero Saynatkari) almost 15 years ago

=begin
Excerpts from Hugh Sasse's message of Mon Jan 25 17:36:06 +0200 2010:

Had a look at Ruby Spec: Seems to require a number of things not
shipped with Ruby, rspec[?] and a Mocking Framework, requires a
github account to be able to contribute.

RubySpec uses MSpec rather than RSpec. It was written as a
part of the project, and is essentially a vastly simpler
codebase. Some additions like platform- and version guards,
tags and interpreter selection are included. I think you
could currently run the specs on RSpec as the syntax is
compatible. Mocking is included in MSpec itself.

In order to have commit access you will need to have an
account on Github, yes. You can, however, submit patches
through the tracker or even just on the mailing list.

Until people get tired of applying them and tell you
to just sign up so you can do it yourself, that is :)

                                 The tests in RubySpec

don't seem to include Regexps with captures, which mine do. I
don't yet know how to rewrite mine to fit Ruby Spec syntax.

These would be an excellent addition, so hopefully you will
be able to. You can ask for help on the ML or on #rubyspec
on Freenode.

 http://rubyspec.org

Eero

Magic is insufficiently advanced technology.

=end

Actions #11

Updated by mg (Dawid Grzesiak) over 14 years ago

=begin
Hi,

I suppose that:
irb(main):050:0> ".".split "."
=> []

will give me an array with two empty strings instead of empty array.
How can I do that?

Moreover:
irb(main):051:0> ".l...".split "."
=> ["", "l"]

Shouldn't it yield ["", "l", "", "", ""] array?

Thanks,
Dawid
=end

Actions #12

Updated by mg (Dawid Grzesiak) over 14 years ago

=begin
Or in this example:

irb(main):053:0> "..l..".split "."
=> ["", "", "l"]

there is no symmetry :/
=end

Actions #13

Updated by hgs (Hugh Sasse) over 14 years ago

=begin
On Thu, 4 Mar 2010, Dawid Grzesiak wrote:

Issue #2647 has been updated by Dawid Grzesiak.

Or in this example:

irb(main):053:0> "..l..".split "."
=> ["", "", "l"]

there is no symmetry :/

hgs@Q2P14HGS ~
12:27:13$ ruby -e 'puts "..l..".split(/./).inspect'
["", "", "l"]
# Confirmed for regexps
hgs@Q2P14HGS ~
12:27:36$ ruby -e 'puts "..l..".split(/./,-1).inspect'
["", "", "l", "", ""]
# Limit parameter solves this
hgs@Q2P14HGS ~
12:27:46$

ri String#split gives:

----------------------------------------------------------- String#split
str.split(pattern=$;, [limit]) => anArray

  From Ruby 1.8


     [...]
  If the _limit_ parameter is omitted, trailing null fields are
  suppressed. If _limit_ is a positive number, at most that number of
  fields will be returned (if _limit_ is +1+, the entire string is
  returned as the only entry in an array). If negative, there is no
  limit to the number of fields returned, and trailing null fields
  are not suppressed.


     HTH
     Hugh

=end

Actions

Also available in: Atom PDF

Like0
Like0Like0Like0Like0Like0Like0Like0Like0Like0Like0Like0Like0Like0