[cocci] Avoiding duplicate elements in SmPL disjunctions?

[cocci] Avoiding duplicate elements in SmPL disjunctions?

[Markus Elfring]

Hello,

I tried the following script variant out for the semantic patch language.

@display@
statement s;
@@
*while (1)
    s


Questionable test result (according to the software combination “Coccinelle 1.3.0”):
Markus_Elfring@Sonne:…/Projekte/Coccinelle/janitor> /usr/bin/spatch --parse-cocci show_unlimited_while_loops.cocci
…
(
*while *(*1*) 
|
*while *(*1 *!= *0*) 
|
*while *(*1*) 
|
*while *(*0 *!= *1*) 
)s
…


An element of the generated SmPL disjunction got repeated somehow.
How do you think about to avoid such duplicate code?

Regards,
Markus

Re: [cocci] Avoiding duplicate elements in SmPL disjunctions?

[Markus Elfring]

> An element of the generated SmPL disjunction got repeated somehow.
> How do you think about to avoid such duplicate code?

Will another data representation trigger more desirable improvements?

Markus_Elfring@Sonne:…/Projekte/Coccinelle/janitor> /usr/bin/spatch --parse-cocci show_unlimited_while_loops.cocci | sort | uniq -c -d
…
      2 *while *(*1*) 
Markus_Elfring@Sonne:~> /usr/bin/spatch --parse-cocci --sp '! 2' | sort | uniq -c -d
…
      3 *0 *== *2
      3 *!*2
      3 *2 *== *0


Which components of the Coccinelle software are responsible for such
data processing results?

Regards,
Markus

Re: [cocci] Avoiding duplicate elements in SmPL disjunctions?

[Markus Elfring]

>> Which components of the Coccinelle software are responsible for such
>> data processing results?
> iso_pattern.ml

Thanks that you pointed this OCaml source file out finally.
https://gitlab.inria.fr/coccinelle/coccinelle/-/blob/2d5df416a939c5c7b9d958559be3b63e0b1414fe/parsing_cocci/iso_pattern.ml

Unfortunately, I find that such source code is not directly adjustable
(by me).
Thus I hope that algorithmic properties can be clarified better
also according to reported remaining open issues.

Regards,
Markus

Re: [cocci] Avoiding duplicate elements in SmPL disjunctions?

[Markus Elfring]

> iso_pattern.ml

Do test results indicate that data processing for isomorphisms is still so erroneous
that this functionality should be deactivated until the implementation quality
will really be acceptable?

How can it be switched off?
https://gitlab.inria.fr/coccinelle/coccinelle/-/blob/2d5df416a939c5c7b9d958559be3b63e0b1414fe/docs/manual/spatch_options.tex#L93-112

Regards,
Markus

Re: [cocci] Avoiding duplicate elements in SmPL disjunctions?

[Markus Elfring]

>> Do test results indicate that data processing for isomorphisms is still so erroneous
>> that this functionality should be deactivated until the implementation quality
>> will really be acceptable?
> 
> The implementation quality is fine.

I got an other impression.


> No one has complained for the last 15 years.

Am I a contributor who presented relevant bug reports?

How will development interests (and resources) evolve further?

Regards,
Markus

Re: [cocci] Avoiding duplicate elements in SmPL disjunctions?

[Markus Elfring]

> Markus_Elfring@Sonne:~> /usr/bin/spatch --parse-cocci --sp '! 2' | sort | uniq -c -d
> …
>       3 *0 *== *2
>       3 *!*2
>       3 *2 *== *0

Can a command like “`spatch --parse-cocci --sp '! 2' | sort | uniq -c -d`”
trigger further test considerations according to Coccinelle 1.3.1 and 1.3.2?

Questionable test result:
…
      5 
      3 @@
      8 |


Regards,
Markus

Re: [cocci] Avoiding duplicate elements in SmPL disjunctions?

[Julia Lawall]

[-- Attachment #1: Type: text/plain, Size: 476 bytes --]



On Tue, 25 Aug 2026, Markus Elfring wrote:

> > Markus_Elfring@Sonne:~> /usr/bin/spatch --parse-cocci --sp '! 2' | sort | uniq -c -d
> > …
> >       3 *0 *== *2
> >       3 *!*2
> >       3 *2 *== *0
>
> Can a command like “`spatch --parse-cocci --sp '! 2' | sort | uniq -c -d`”
> trigger further test considerations according to Coccinelle 1.3.1 and 1.3.2?
>
> Questionable test result:
> …
>       5
>       3 @@
>       8 |

I have no idea what this means.

julia

Re: [cocci] Avoiding duplicate elements in SmPL disjunctions?

[Markus Elfring]

>> Can a command like “`spatch --parse-cocci --sp '! 2' | sort | uniq -c -d`”
>> trigger further test considerations according to Coccinelle 1.3.1 and 1.3.2?
>>
>> Questionable test result:
>> …
>>       5
>>       3 @@
>>       8 |
> 
> I have no idea what this means.
Would any more software users dare to retry mentioned test commands?

Regards,
Markus