Add support for new absence operator (#33)

[jaynetics]

No description provided.

[ammar]

Hi @Janosch-x . Thanks for this and sorry about the delayed response.

My initial thinking was this new "operator" is not really a "group" and thus had started investigating an alternative approach. However the fact that the new construct allows for any sub-expression to appear as a exp makes using the group approach easier because it will use the existing scanner recursion. for example in a case like (?~abc|def|[ghi])

So, this is probably the best approach now and it looks good.

Cheers and thanks again!

[ammar]

@Janosch-x

On further review I noticed a couple of things:

• Need to double check the addition of Absent to Group::All based on where it is used (see ruby 1.8.6 syntax file)
• The new type/token (:group/:absent) should be added to the ruby 2.4.1 syntax file with a call to implements.

[jaynetics]

Hi @ammar,

thanks for the feedback.

I also pondered whether it really is a group, ending up with the same conclusion as you did. To be sure I even asked the author, and he does see it is a group :)

Need to double check the addition of Absent to Group::All based on where it is used (see ruby 1.8.6 syntax file)

This is indeed a mistake. Do you have a preferred solution? I think All should contain Absent, because it doesn't seem right to handle version differences in Token::Group. Maybe we could write Group::All - [Group::Absent] in the 1.8.6 syntax file? Of course that is not very pretty either ...

[jaynetics]

Or maybe remove Group::All and list all the "classical" group types individually in the 1.8.6 file?

[jaynetics]

@ammar In accordance with other token files I've removed the new type constant from the All array and called implement on it separately in the 2.4.1 file.

There is also a syntax file test and a parser test for it now.

Lastly, to match the naming in Onigmo, I have changed the name from "absent" to "absence".

[ammar]

@Janosch-x Sorry for not responding sooner. I recently had an accident and have an arm and a shoulder in cast. Hope to be able to work again soon

[jaynetics]

@ammar I'm sorry to hear that. Please don't feel pressurised by my landslide of PRs. There is no urgency there at all, I had just been bored. Wishing you a good recovery!

[ammar]

Hi @Janosch-x, thanks for looking into the classification of the operator and the updates. Again, I'm very sorry for the very late response.

I think this looks good and I would like to merge it.

Cheers!

[ammar]

Included in release 0.4.4