[Cocci] Working with backreferences in SmPL scripts?

[Cocci] Working with backreferences in SmPL scripts?

[Markus Elfring]

Hello,

I have tried another small SmPL script out.


@replacement@
expression ex;
identifier var;
@@
 var
-   = var +
+   +=
            ex


elfring@Sonne:~/Projekte/Coccinelle/Probe> spatch simplify_addition1.cocci Test_increment1.c
…
@@ -1,6 +1,6 @@
 int main(void)
 {
 int counter = 3;
-counter = counter + 5;
+counter += 5;
 return counter;
 }


I have repeated this transformation approach with the following source code.


int main(void)
{
int x = 2, y = 3;
x = y + 5;
return x;
}


This test example did not present a generated change suggestion.
I find that such a result can be interpreted in two ways then.

1. The Coccinelle software ensured that the specified metavariable “var”
   referred to the same source code at all shown places with the consequence
   that the variable “y” could not match in this test case.

2. A possible data output was not displayed despite of the detail
   that the variable “y” is also an identifier according to the syntax of
   the C programming language.
   Would there be a need to express a relationship to a previous match
   by a kind of notation for backreferences?


The situation can become challenging for the safe handling of advanced
statements in more meaningful source files.

* How do you think about to help with the reconsideration
  (and occasional deletion) of duplicate code by the means of
  the semantic patch language?

* Would you like to clarify the desired software behaviour for such use cases?

Regards,
Markus
_______________________________________________
Cocci mailing list
Cocci@systeme.lip6.fr
https://systeme.lip6.fr/mailman/listinfo/cocci

Re: [Cocci] Working with backreferences in SmPL scripts?

[Julia Lawall]

[-- Attachment #1: Type: text/plain, Size: 1954 bytes --]



On Wed, 17 Apr 2019, Markus Elfring wrote:

> Hello,
>
> I have tried another small SmPL script out.
>
>
> @replacement@
> expression ex;
> identifier var;
> @@
>  var
> -   = var +
> +   +=
>             ex
>
>
> elfring@Sonne:~/Projekte/Coccinelle/Probe> spatch simplify_addition1.cocci Test_increment1.c
> …
> @@ -1,6 +1,6 @@
>  int main(void)
>  {
>  int counter = 3;
> -counter = counter + 5;
> +counter += 5;
>  return counter;
>  }
>
>
> I have repeated this transformation approach with the following source code.
>
>
> int main(void)
> {
> int x = 2, y = 3;
> x = y + 5;
> return x;
> }

A metavariable is always bound to the same value within a given
control-flow path.  I have no idea what you are trying to get at with
this example.

julia

>
>
> This test example did not present a generated change suggestion.
> I find that such a result can be interpreted in two ways then.
>
> 1. The Coccinelle software ensured that the specified metavariable “var”
>    referred to the same source code at all shown places with the consequence
>    that the variable “y” could not match in this test case.
>
> 2. A possible data output was not displayed despite of the detail
>    that the variable “y” is also an identifier according to the syntax of
>    the C programming language.
>    Would there be a need to express a relationship to a previous match
>    by a kind of notation for backreferences?
>
>
> The situation can become challenging for the safe handling of advanced
> statements in more meaningful source files.
>
> * How do you think about to help with the reconsideration
>   (and occasional deletion) of duplicate code by the means of
>   the semantic patch language?
>
> * Would you like to clarify the desired software behaviour for such use cases?
>
> Regards,
> Markus
> _______________________________________________
> Cocci mailing list
> Cocci@systeme.lip6.fr
> https://systeme.lip6.fr/mailman/listinfo/cocci
>

[-- Attachment #2: Type: text/plain, Size: 136 bytes --]

_______________________________________________
Cocci mailing list
Cocci@systeme.lip6.fr
https://systeme.lip6.fr/mailman/listinfo/cocci

Re: [Cocci] Working with backreferences in SmPL scripts?

[Markus Elfring]

> A metavariable is always bound to the same value within a given
> control-flow path.

Thanks for this information.


> I have no idea what you are trying to get at with this example.

Some goals can match to my use case description.

* Is another nice refactoring application demonstrated for
  the semantic patch language to trigger constructive development discussions?

* Can such test cases help to clarify views about questionable software behaviour
  (from analysis of longer source files with more interesting statements)?

Regards,
Markus
_______________________________________________
Cocci mailing list
Cocci@systeme.lip6.fr
https://systeme.lip6.fr/mailman/listinfo/cocci