[erlang-questions] erlang-questions] inets httpd mod_log output issue

Ingela Anderton Andin <>
Mon May 12 11:59:48 CEST 2008


Paul Fisher wrote:
> On Mon, 2008-05-12 at 10:39 +0200, Ingela Anderton Andin wrote:
>   
>>> The 'compact' 'error_log_format' generates things like the following in
>>> the
>>> log files:
>>>
>>> [09/May/2008:09:19:57 -0500] access to /foo/flooga.txt failed for 127.0.0.1, reason: >[104,116,116,112,100,95,102,105,108,101,58,32,67,97,110,39,116,32,111,112,101,110,47,104,111,109,101,47,112,102,105,115,104,101,114,47,108,109,47,108,109,45,116,114,117,110,107,47,98,108,100,47,101,114,108,97,110,103,47,105,110,101,116,115,109,103,114,47,112,114,105,118,47,102,111,111,47,102,108,111,111,103,97,46,116,120,116]
>>>
>>>
>>>       
>> Yes we know. Maybe you should be using pretty insteadof compact. The assumption here is that
>> if you want to use a tool to read the logs it will be very nice that each entry is only one line. And if
>> you are using a tool it will be easy to let the tool do its own formatting of the entry. But 
>> if the logs are mainly read by humans directly you should use pretty format. 
>>     
>
> We (like I suspect lots of other production shops) write human
> readable/parsible logs as single line per entry, which is what 'compact'
> seemed to be.
>
>   
Well, if that is what web-application  developers wants maybe that is 
what we should make it into?!
Maybe our first assumption does not hold up.
I am just trying to explain what the initial intent was. We will take 
this under consideration.

> Why support the 'compact' form if it generates "write-only" log messages
> (intended never to be read by human or machine)?  Seriously, if there is
> no intent to "fix" the 'compact' form to print intelligible information,
> then deprecate it.
>   
It was intended  to be read  by a tool and serve as some kind of 
backwards compatibility
when we made the pretty format.

Regards Ingela, Erlang/OTP Ericsson

 





More information about the erlang-questions mailing list