The Inbox: Regex-Core-ul.57.mcz

Am 12.04.2020 09:29:27 schrieb Levente Uzonyi <[hidden email]>:

Hi Christoph,

On Sat, 11 Apr 2020, Thiede, Christoph wrote:

>
> Great idea! Are there any visible performance improvements? In any case, this change makes it more easy to explore and debug the matching process!

There's ~5% speedup with the simple test case I was benchmarking.

>
>
> FYI, you left three return carets in this method. :-)

Right. I wanted to move the return instruction outside but forgot to
remove the inner ones.


Levente

>
>
> Best,
> Christoph
>
> __________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________
> Von: Squeak-dev im Auftrag von [hidden email]
> Gesendet: Samstag, 11. April 2020 22:05:05
> An: [hidden email]
> Betreff: [squeak-dev] The Inbox: Regex-Core-ul.57.mcz  
> Levente Uzonyi uploaded a new version of Regex-Core to project The Inbox:
> http://source.squeak.org/inbox/Regex-Core-ul.57.mcz
>
> ==================== Summary ====================
>
> Name: Regex-Core-ul.57
> Author: ul
> Time: 11 April 2020, 9:42:43.405809 pm
> UUID: 3cda3ad5-49f2-41f1-ab9d-2f6fc242ae9d
> Ancestors: Regex-Core-ct.56
>
> - avoid generating RxmBranch nodes when there's no real branch. e.g. 'abc' asRegex
>
> =============== Diff against Regex-Core-ct.56 ===============
>
> Item was changed:
>   ----- Method: RxMatcher>>hookBranchOf:onto: (in category 'private') -----
>   hookBranchOf: regexNode onto: endMarker
>          "Private - Recurse down the chain of regexes starting at
>          regexNode, compiling their branches and hooking their tails
>          to the endMarker node."
>  
> +        ^regexNode regex
> +                ifNil: [ "Avoid creating a branch without an alternative."
> +                        ^(regexNode branch dispatchTo: self)
> +                                pointTailTo: endMarker;
> +                                yourself ]
> +                ifNotNil: [ :regex |
> +                        | rest |
> +                        rest := self hookBranchOf: regex onto: endMarker.
> +                        ^RxmBranch new
> +                                next: ((regexNode branch dispatchTo: self)
> -        | rest |
> -        rest := regexNode regex ifNotNil: [ :regex |
> -                self hookBranchOf: regex onto: endMarker ].
> -        ^RxmBranch new
> -                next: ((regexNode branch dispatchTo: self)
>                                          pointTailTo: endMarker;
>                                          yourself);
> +                                alternative: rest;
> +                                yourself ]
> + !
> -                alternative: rest;
> -                yourself!
>
>
>
>