Deprecate usage of should_not raise_error

[andrykonchin]

Print deprecation message for every example of -> {}.should_not raise_error in specs. I believe it's enough to ensure that new specs don't contain deprecated DSL.

Related issues:

• Deprecate should_not raise_error in MSpec and replaces usages in ruby/spec spec#754

Example

bin/mspec ../spec/core/proc/curry_spec.rb
$ ruby /Users/andrykonchin/projects/mspec/bin/mspec-run ../spec/core/proc/curry_spec.rb
ruby 2.7.1p83 (2020-03-31 revision a0c7c23c9c) [x86_64-darwin18]
[/ |                   0%                     | 00:00:00]      0F      0E
Deprecation: {}.should_not raise_error breaks code style and is deprecated

Deprecation: {}.should_not raise_error breaks code style and is deprecated
[\ | ==================100%================== | 00:00:00]      0F      0E

Finished in 0.000678 seconds

1 file, 1 example, 2 expectations, 0 failures, 0 errors, 0 tagged

[andrykonchin]

It prevents running MSpec specs on Windows. Not sure whether it's important.

It would be a little more difficult to check STDERR of the command in this test. Another option is to write a deprecation message to STDOUT instead of STDERR.

[eregon]

I think 2>&1 might actually work on Windows, could you move it to the line below though, so file is still an actual file?

[eregon]

We should show the caller here, so it's easy to know where it comes from.

[eregon]

Could you use MSpec.deprecate (

mspec/lib/mspec/utils/deprecate.rb

Line 2 in 47e3cbf

def self.deprecate(what, replacement)

)?
That will automatically include the caller.

[eregon]

replacement can be e.g. "a matcher to verify the result"

[andrykonchin]

Yeah, will try. I saw this helper but decided not to use it because it has a little bit different purpose - to notify about changing in API or DSL. But here we have a code style issue.

Nevertheless it should work well for our case.

[andrykonchin]

Done

[eregon]

I think 2>&1 might actually work on Windows, could you move it to the line below though, so file is still an actual file?

[eregon]

I think we should start a PR on ruby/spec to fix the warnings, because if we merge and there are hundreds of warnings it will be very annoying for all people running ruby/spec.

[andrykonchin]

There are about 180 examples to change. Will try to fix them soon.