Another minor =~ glitch

Another minor =~ glitch

[Bart Schaefer]

Side note to the REMATCH_PCRE thread:  xtrace output shows what's being
done, but it can't be "played back":

torch% set -x
torch% [[ foo =~ .* ]]
+Src/zsh:2> [[ foo -regex-match .* ]]
torch% [[ foo -regex-match .* ]]
+Src/zsh:3> [[zsh: unknown condition: -regex-match
 ]]
torch% 

Re: Another minor =~ glitch

[Roland Eggner]

On 2017-04-07 Fr 10:33:20-0700, Bart Schaefer wrote:
> torch% set -x
> torch% [[ foo =~ .* ]]
> +Src/zsh:2> [[ foo -regex-match .* ]]
> torch% [[ foo -regex-match .* ]]
> +Src/zsh:3> [[zsh: unknown condition: -regex-match
>  ]]
> torch% 

Regression observed since my Zsh update
from bb6c08b51a079870 to fc1fedda954e1d10.

Similarly using operator “-pcre-match” fails,
using operator “=~” instead works around:

syst% zmodload zsh/pcre
syst% zmodload -F -l zsh/pcre
+b:pcre_compile
+b:pcre_match
+b:pcre_study
+C:pcre-match
syst% setopt RE_MATCH_PCRE XTRACE
syst% [[ "A" -pcre-match "(?i)a" ]]
[[zsh: unknown condition: -pcre-match
 ]]
syst% [[ "A" =~ "(?i)a" ]]  && print y
[[ "A" -pcre-match "(?i)a" ]]
print y
y
syst%



-- 
Roland Eggner

Re: Another minor =~ glitch

[Bart Schaefer]

On Fri, Apr 7, 2017 at 6:58 PM, Roland Eggner <edvz2@systemanalysen.net> wrote:
>
> Regression observed since my Zsh update
> from bb6c08b51a079870 to fc1fedda954e1d10.

Thanks, that narrows it down to this:

diff --git a/Src/cond.c b/Src/cond.c
index 9b739f6..a638412 100644
--- a/Src/cond.c
+++ b/Src/cond.c
@@ -139,9 +139,9 @@ evalcond(Estate state, char *fromtest)
                l = 2;
            }
            if (name && IS_DASH(name[0]))
-               errname = name;
+               untokenize(errname = name);
            else if (strs[0] && IS_DASH(*strs[0]))
-               errname = strs[0];
+               untokenize(errname = strs[0]);
            else
                errname = "<null>";
            if (name && IS_DASH(name[0]) &&

Re: Another minor =~ glitch

[Mikael Magnusson]

On Sat, Apr 8, 2017 at 7:23 AM, Bart Schaefer <schaefer@brasslantern.com> wrote:
> On Fri, Apr 7, 2017 at 6:58 PM, Roland Eggner <edvz2@systemanalysen.net> wrote:
>>
>> Regression observed since my Zsh update
>> from bb6c08b51a079870 to fc1fedda954e1d10.
>
> Thanks, that narrows it down to this:
>
> diff --git a/Src/cond.c b/Src/cond.c
> index 9b739f6..a638412 100644
> --- a/Src/cond.c
> +++ b/Src/cond.c
> @@ -139,9 +139,9 @@ evalcond(Estate state, char *fromtest)
>                 l = 2;
>             }
>             if (name && IS_DASH(name[0]))
> -               errname = name;
> +               untokenize(errname = name);
>             else if (strs[0] && IS_DASH(*strs[0]))
> -               errname = strs[0];
> +               untokenize(errname = strs[0]);
>             else
>                 errname = "<null>";
>             if (name && IS_DASH(name[0]) &&

This patch causes segfaults for me (in the first hunk at least)
because name can be a readonly string. Should we ztrdup it?

-- 
Mikael Magnusson

Re: Another minor =~ glitch

[Mikael Magnusson]

On Wed, Jun 14, 2017 at 6:45 AM, Mikael Magnusson <mikachu@gmail.com> wrote:
> On Sat, Apr 8, 2017 at 7:23 AM, Bart Schaefer <schaefer@brasslantern.com> wrote:
>> On Fri, Apr 7, 2017 at 6:58 PM, Roland Eggner <edvz2@systemanalysen.net> wrote:
>>>
>>> Regression observed since my Zsh update
>>> from bb6c08b51a079870 to fc1fedda954e1d10.
>>
>> Thanks, that narrows it down to this:
>>
>> diff --git a/Src/cond.c b/Src/cond.c
>> index 9b739f6..a638412 100644
>> --- a/Src/cond.c
>> +++ b/Src/cond.c
>> @@ -139,9 +139,9 @@ evalcond(Estate state, char *fromtest)
>>                 l = 2;
>>             }
>>             if (name && IS_DASH(name[0]))
>> -               errname = name;
>> +               untokenize(errname = name);
>>             else if (strs[0] && IS_DASH(*strs[0]))
>> -               errname = strs[0];
>> +               untokenize(errname = strs[0]);
>>             else
>>                 errname = "<null>";
>>             if (name && IS_DASH(name[0]) &&
>
> This patch causes segfaults for me (in the first hunk at least)
> because name can be a readonly string. Should we ztrdup it?

(I probably meant dupstring)

-- 
Mikael Magnusson

PATCH: dupstring a possibly readonly string before modifying

[Mikael Magnusson]

---
 Src/cond.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/Src/cond.c b/Src/cond.c
index a63841234d..b9a47cea54 100644
--- a/Src/cond.c
+++ b/Src/cond.c
@@ -139,7 +139,7 @@ evalcond(Estate state, char *fromtest)
 		l = 2;
 	    }
 	    if (name && IS_DASH(name[0]))
-		untokenize(errname = name);
+		untokenize(errname = dupstring(name));
 	    else if (strs[0] && IS_DASH(*strs[0]))
 		untokenize(errname = strs[0]);
 	    else
-- 
2.8.2