Add Int#bits for accessing bit ranges

[stakach]

inspired by https://twitter.com/iximeow/status/1170530444112097282?s=21

[stakach]

Weird... I don't see the overflow running on my Mac.
Any hints on how I should ignore that?

[asterite]

In the past we decided to explicitly not add this. Imagine passing an integer, or an array size, or whatever number to a method that you expect is something that indexes an array. Then your code compiles and you have no idea why. It can be really confusing and I experienced that in Ruby. We have the bit method for this.

[stakach]

Good point!
What about a #bits function?

[asterite]

What about a #bits function?

Maybe :-)

[RX14]

Specs fail cause of overflow.

[stakach]

@RX14 Yeah, any pointers on how to ignore that?

[bcardiff]

@stakach the overflow is in the new code: https://github.com/crystal-lang/crystal/pull/8165/files#diff-cf769fe2e361e14888004d0eae8368fdR366

Try building with -Dpreview_overflow and it should raise on your side aswell.

If you need wrapping around operations for + - *, use &+, &-, &*.

[asterite]

The code looks code, the method is fancy and we could merge it. I just like to know what use cases this has.

[stakach]

It's useful for dealing with binary data structures.
https://en.wikipedia.org/wiki/IPv4#/media/File:Ipv4_address.svg

IPv4 is probably not a great example, but I've had to do things like extract the components of a custom floating point number format. So many weird and wonderful protocols out there where this is more descriptive and less error prone than bit shifting and masking.

i.e. https://github.com/lifeemotions/knx.net/blob/master/src/KNXLib/DPT/DataPoint2ByteFloatTemperature.cs#L28
How ugly is that code

[refi64]

Would it be worth changing the PR title prior to merging to avoid confusion as to what was actually added?

…

On Thu, Sep 19, 2019, 5:41 PM Stephen von Takach ***@***.***> wrote: It's useful for dealing with binary data structures. https://en.wikipedia.org/wiki/IPv4#/media/File:Ipv4_address.svg IPv4 is probably not a great example, but I've had to do things like extracts the components of a custom floating point number format. So many weird and wonderful protocols out there where this is more descriptive and less error prone than bit shifting and masking. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#8165?email_source=notifications&email_token=AAM4YSKWQK6WBBR3KNM4WZLQKP5Z3A5CNFSM4IUYAOIKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD7FA2AA#issuecomment-533335296>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAM4YSL5T7WMU3G67VZVKIDQKP5Z3ANCNFSM4IUYAOIA> .

[icy-arctic-fox]

Ruby 2.7 now has the original syntax used in this PR (without the #bits method). https://blog.saeloun.com/2019/09/24/ruby-2-7-integer-with-range.html

[stakach]

I am so glad to be programming crystal as my primary language:
ruby/ruby@6bedbf4

[sdogruyol]

Thank you @stakach 🙏