> I found the difference of the behavior between Upstream Ruby and Fedora Ruby.
> Case 2-1. does not install ri document by "gem install".

When I tested some things for current private-ruby-2.5 branch, the
result is okay like this.

2-1. Fedora Ruby by root user

```
<mock-chroot> sh-4.4# gem install webrick
Fetching: webrick-1.4.1.gem (100%)
Successfully installed webrick-1.4.1
Parsing documentation for webrick-1.4.1
Installing ri documentation for webrick-1.4.1
Done installing documentation for webrick after 0 seconds
1 gem installed
```

2-2. Fedora Ruby by regular user

```
[mockbuild@23d5fd90050e4ad382984b434162aa67 ~]$ gem install webrick
Fetching: webrick-1.4.1.gem (100%)
Successfully installed webrick-1.4.1
Parsing documentation for webrick-1.4.1
Installing ri documentation for webrick-1.4.1
Done installing documentation for webrick after 0 seconds
1 gem installed
```

On Tue, Dec 19, 2017 at 7:24 PM, Jun Aruga <[email protected]> wrote:
> One more thing.
>
> You would find format errors for your modification part if you run `rubocop`.
>
> ```
> $ rubocop operating_system.rb
> ```
>
> Jun
>
>
> On Tue, Dec 19, 2017 at 7:19 PM, Jun Aruga <[email protected]> wrote:
>> Vit thanks for the working.
>>
>> I tested it.
>>
>>> 1) If "gem install" as a regular user still works the same.
>>
>>
>> Right now there is a 2 type of packages.
>> Some gem package are "default", others are not.
>> The gem package that is managed the ruby sub package is not "default"
>> like "bigdecimal"
>>
>> What is the future plan for these kind of pacakge such as "bigdecimal"?
>> 1. These are "default" removing the sub package?
>> Or 2. Current default package like "cmath" becomes not "default"
>> creating the sub package?
>>
>>
>> Fedora Ruby by regular user.
>>
>>
>> ```
>> [mockbuild@026f2c75e4664cfe887005e710a5497e ~]$ gem list | grep default
>> cmath (default: 1.0.0)
>> csv (default: 1.0.0)
>> date (default: 1.0.0)
>> dbm (default: 1.0.0)
>> digest (default: 0.1.0)
>> etc (default: 1.0.0)
>> fcntl (default: 1.0.0)
>> fiddle (default: 1.0.0)
>> fileutils (default: 1.0.1)
>> gdbm (default: 2.0.0)
>> ipaddr (default: 1.2.0)
>> scanf (default: 1.0.0)
>> sdbm (default: 1.0.0)
>> stringio (default: 0.0.1)
>> strscan (default: 0.0.1)
>> webrick (default: 1.4.0.beta1)
>> zlib (default: 1.0.0)
>>
>> [mockbuild@026f2c75e4664cfe887005e710a5497e ~]$ gem list | grep -v default
>> bigdecimal (1.3.3)
>> did_you_mean (1.1.2)
>> io-console (0.4.6)
>> json (2.1.0)
>> minitest (5.10.3)
>> net-telnet (0.1.1)
>> openssl (2.1.0.beta2)
>> power_assert (1.1.1)
>> psych (3.0.0)
>> rake (12.3.0)
>> rdoc (6.0.0)
>> test-unit (3.2.7)
>> xmlrpc (0.3.0)
>> ```
>>
>> On upstream Ruby
>>
>> ```
>> $ dest/bin/gem list
>>
>> *** LOCAL GEMS ***
>>
>> bigdecimal (default: 1.3.3)
>> bundler (default: 1.16.1.pre1)
>> cmath (default: 1.0.0)
>> csv (default: 1.0.0)
>> date (default: 1.0.0)
>> dbm (default: 1.0.0)
>> digest (default: 0.1.0)
>> etc (default: 1.0.0)
>> fcntl (default: 1.0.0)
>> fileutils (default: 1.0.1)
>> gdbm (default: 2.0.0)
>> io-console (default: 0.4.6)
>> ipaddr (default: 1.2.0)
>> json (default: 2.1.0)
>> openssl (default: 2.1.0)
>> psych (default: 3.0.0)
>> rdoc (default: 6.0.0)
>> scanf (default: 1.0.0)
>> sdbm (default: 1.0.0)
>> stringio (default: 0.0.1)
>> strscan (default: 0.0.1)
>> webrick (default: 1.4.0.beta1)
>> zlib (default: 1.0.0)
>> ```
>>
>>> 2) If "gem install" as root still works the same.
>>
>> "gem list" result is same for regular user's situation.
>>
>> I found the difference of the behavior between Upstream Ruby and Fedora Ruby.
>> Case 2-1. does not install ri document by "gem install".
>>
>> 1-1. Upstream Ruby by root user
>>
>> [root@unused-4-164 ~]# /usr/local/ruby-2.5.0.pre1/bin/gem install digest
>> Fetching: digest-0.0.1.gem (100%)
>> Successfully installed digest-0.0.1
>> Parsing documentation for digest-0.0.1
>> Installing ri documentation for digest-0.0.1
>> Done installing documentation for digest after 0 seconds
>> 1 gem installed
>>
>> 2-1. Fedora Ruby by root user
>>
>> <mock-chroot> sh-4.4# gem install webrick
>> Fetching: webrick-1.4.1.gem (100%)
>> Successfully installed webrick-1.4.1
>> 1 gem installed
>>
>> 2-2. Fedora Ruby by regular user
>>
>> [mockbuild@c187f3581b4e45ecb2837fe5ab6a0af5 ~]$ gem install webrick
>> Fetching: webrick-1.4.1.gem (100%)
>> WARNING:  You don't have /builddir/bin in your PATH,
>>     gem executables will not run.
>> Successfully installed webrick-1.4.1
>> Parsing documentation for webrick-1.4.1
>> Installing ri documentation for webrick-1.4.1
>> Done installing documentation for webrick after 0 seconds
>> 1 gem installed
>>
>>
>> I do not know this difference is this Fedora Ruby specific.
>> This might be related to this issue?
>> https://src.fedoraproject.org/rpms/ruby/pull-request/9
>>
>>> 3) If the RPM packages in Fedora (probably just noarch) still installs
>> and runs just fine.
>>> 4) If rubygem- RPM packages build using this ruby are still build and
>>
>> I could not test below cases. I tried to build rubygem-bundler for
>> your Ruby RPMs.
>> But I could not build because of conflict with ruby-2.4.2.
>>
>> rubygem-bundler
>>
>> ```
>> $ mock -r fedora-rawhide-x86_64 --with tests -n *.rpm
>> ...
>> Error:
>>  Problem: cannot install both ruby-libs-2.4.2-85.fc28.x86_64 and
>> ruby-libs-2.5.0-0.1.r61214.fc28.x86_64
>> ```
>>
>>
>> Jun
>>
>> On Tue, Dec 19, 2017 at 4:12 PM, Vít Ondruch <[email protected]> wrote:
>>>
>>>
>>> Dne 14.12.2017 v 19:03 Vít Ondruch napsal(a):
>>>>
>>>> Dne 14.12.2017 v 18:41 Vít Ondruch napsal(a):
>>>>> Dne 14.12.2017 v 18:23 Jun Aruga napsal(a):
>>>>>> OK thanks for the info.
>>>>>>
>>>>>> Comparing the result of "gem list" command between upstream and our
>>>>>> Fedora package, I found the difference.
>>>>>> That can be confusing people.
>>>>>>
>>>>>> Some of the gem are not shown in the result such as cmath for Fedora
>>>>>> package ruby.
>>>>>>
>>>>>> When running below command on mock, we can load cmath that is not in
>>>>>> "gem list" on mock, maybe those are only shown as a result of "gem
>>>>>> list".
>>>>>>
>>>>>> ```
>>>>>> irb(main):003:0> require 'cmath'
>>>>>> => true
>>>>>> ```
>>>>>>
>>>>>> Is it possible to add those gems in the result as a compatibility for
>>>>>> upstream Ruby?
>>>>>> Hidden gems such as cmath are confusing users.
>>>>> Interesting. That is definitely unintentional. Will take a look into it.
>>>>>
>>>> This appears to be related to the default location of where the gems are
>>>> installed. Upstream Ruby installs the gems into their directory, we
>>>> install the gems into home directory. And therefore RubyGems on Fedora
>>>> are trying to load the specifications for the default gems from the home
>>>> directory "/builddir/.gem/ruby/specifications/default" (testing in
>>>> mock). So far, we never had the default gem specifications, so this was
>>>> not issue.
>>>>
>>>>
>>>
>>>
>>> Here is updated build, which should fix the issues:
>>>
>>> https://koji.fedoraproject.org/koji/taskinfo?taskID=23793602
>>>
>>> The patch used to fix this in attachment. I'd love some feedback prior I
>>> push this into git. Mainly, I'd like you to test:
>>>
>>> 1) If "gem install" as a regular user still works the same.
>>> 2) If "gem install" as root still works the same.
>>> 3) If the RPM packages in Fedora (probably just noarch) still installs
>>> and runs just fine.
>>> 4) If rubygem- RPM packages build using this ruby are still build and
>>> installed correctly.
>>> 5) Any additional scenario you can think of ...
>>>
>>> Thx for testing.
>>>
>>>
>>> Vít
>>>
>>> _______________________________________________
>>> ruby-sig mailing list -- [email protected]
>>> To unsubscribe send an email to [email protected]
>>>
>>
>>
>>
>> --
>> Jun Aruga [email protected]
>> IRC: jaruga, Office: TPB(Technology Park Brno) Building C 1F, Brno,
>> Czech Republic
>
>
>
> --
> Jun Aruga [email protected]
> IRC: jaruga, Office: TPB(Technology Park Brno) Building C 1F, Brno,
> Czech Republic



-- 
Jun Aruga [email protected]
IRC: jaruga, Office: TPB(Technology Park Brno) Building C 1F, Brno,
Czech Republic
_______________________________________________
ruby-sig mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to