On 8/14/24 7:19 AM, Steffen wrote:
> 
> 
>> Op 13 aug 2024 om 10:26 heeft Steffen <i...@apachelounge.com> het volgende 
>> geschreven:
>>
>> 
>>
>>
>>> Op 13 aug 2024 om 10:00 heeft Ruediger Pluem <rpl...@apache.org> het 
>>> volgende geschreven:
>>>
>>> 
>>>
>>> On 8/13/24 9:56 AM, SteffenAL wrote:
>>>>  
>>>>  
>>>> On Tuesday 13/08/2024 at 08:54, Ruediger Pluem wrote:
>>>>>
>>>>>
>>>>> On 8/12/24 5:44 PM, Eric Covener wrote:
>>>>>> On Mon, Aug 12, 2024 at 5:18 AM Ruediger Pluem <rpl...@apache.org> wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On 8/7/24 2:37 PM, Ruediger Pluem wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>> On 7/30/24 9:52 AM, SteffenAL wrote:
>>>>>>>>>
>>>>>>>>> Please have a look
>>>>>>>>>
>>>>>>>>> https://www.apachelounge.com/viewtopic.php?p=42911
>>>>>>>>
>>>>>>>> Can we have the full LogLevel trace8 error_log posted here or in the 
>>>>>>>> forum (the link above only contains a snipped).
>>>>>>>
>>>>>>> Meanwhile I got the logs off-list. Thanks. Can you please add the UNC 
>>>>>>> flag
>>>>>>> (https://httpd.apache.org/docs/2.4/rewrite/flags.html#flag_unc) to each 
>>>>>>> of your .htaccess / per directory rewriterules?
>>>>>>>
>>>>>>> To my fellow developers:
>>>>>>>
>>>>>>> I think the slash merging block in case of a per-dir rewrite is wrong:
>>>>>>>
>>>>>>>       if (!(p->flags & RULEFLAG_UNC)) {
>>>>>>>           /* merge leading slashes, unless they were literals in the 
>>>>>>> sub */
>>>>>>>           if (!AP_IS_SLASH(p->output[0]) || !AP_IS_SLASH(p->output[1])) 
>>>>>>> {
>>>>>>>               while (AP_IS_SLASH(r->filename[0]) &&
>>>>>>>                      AP_IS_SLASH(r->filename[1])) {
>>>>>>>                   r->filename++;
>>>>>>>               }
>>>>>>>           }
>>>>>>>       }
>>>>>>>
>>>>>>> The dir prefix which is already added at this point of time can start 
>>>>>>> with '//' while the output of the rule does not as it is
>>>>>>> relative to the directory. I think in the per-dir context this check is 
>>>>>>> not needed and should be removed.
>>>>>>
>>>>>> Good find. It seems like the substitution itself could still be
>>>>>> tricked into // in perdir context though, and then no prefix is added.
>>>>>> So I guess we should track directly whether any prefix was added as in
>>>>>> https://people.apache.org/~covener/patches/rewrite-unc-perdir.diff ?
>>>>>>
>>>>>
>>>>> +1. Can you commit?
>>>>>
>>>>> @SteffenAl: Is there a way for the reporter in the forum to test the 
>>>>> patch from Eric?
>>>>>
>>>>> Regards
>>>>>
>>>>> Rüdiger
>>>>
>>>> Done
>>>
>>> Thanks
>>>
>>> Regards
>>>
>>> Rüdiger
>>>
>> He is out of office until Thursday August 15th
> 
> Good news : 
> 
> Confirmed - using the provided test version of mod_rewrite allows the site to 
> load and function as expected, without errors.
> 
> The UNC flag was not added to any of the site .htaccess files for the test.
> 

Great. Eric committed as r1919860 to trunk and already proposed for backport.

Regards

Rüdiger

Reply via email to