Re: Using XML for audit trail

OK, thanks for the clarification.  It sounds like by using fragments I
should be safe and I can avoid creating my own record format (which is why
we have XML in the first place).
-Mont


On Fri, Feb 27, 2009 at 1:59 PM, John Brett <[email protected]> wrote:

> Perhaps I should explain. The server received fragments of text which is
> appended to and XML file. Sometimes the server hung and the log file was
> there for diagnosis but because and incomplete fragment had been written out
> the file wasn't valid XML.
>
> I wish I'd known about XML fragment documents at the time.
>
> Sent from my iPhone
>
>
> On 27 Feb 2009, at 18:19, Mont Rothstein <[email protected]> wrote:
>
>  John,
>>
>> Thanks for the feedback and info.
>>
>> Do you have any insight into what was causing the failures?  Our logs are
>> unlikely to be anything close to 10 MB in size.
>>
>> Thanks,
>> -Mont
>>
>>
>> On Thu, Feb 26, 2009 at 11:41 PM, John Brett <[email protected]
>> >wrote:
>>
>>  Hi Mont,
>>>
>>> I've just been through the same thought exercise.
>>>
>>> I concluded that xml simply isn't a good format to use for a continual
>>> logging system.
>>> My conclusions were, in part, based upon seeing an internal system
>>> that did pretty much what you're describing fail periodically, leaving
>>> behind a log file that was no longer valid xml (and therefore couldn't
>>> be parsed by all the standard tools we had in place). Trying to read a
>>> 10Mb sort-of xml document in a text editor to diagnose faults is
>>> pretty tedious.
>>>
>>> My solution was to use a different serialisation format (structured
>>> text) that was record-oriented, and then to have a separate phase
>>> which could convert a given snapshot of the log into xml.
>>>
>>> HTH,
>>> John
>>>
>>> 2009/2/27 Mont Rothstein <[email protected]>:
>>>
>>>> I want to use XML for an audit trail.  The problem I am having is that I
>>>> would like to only use append operations to write to the file.
>>>>
>>>> My thought had been to write an element to each line, with sub-elements
>>>>
>>> and
>>>
>>>> the end on the same line.  I could then parse each line when reading the
>>>> file.  The problem is I can't figure out how to get the XML for a
>>>> node/element/document, and all of the sub-elements, to write out as a
>>>>
>>> single
>>>
>>>> line.
>>>>
>>>> Alternatively  I could use some other mechanism to parse the file and
>>>>
>>> read
>>>
>>>> each audit entry (xml element with sub-elements) in sequence.
>>>>
>>>> Any suggestions on how to either write the XML as a single line or a
>>>> preferred method for parsing elements in a file one at a time?
>>>>
>>>> Thanks,
>>>> -Mont
>>>>
>>>> ===================================
>>>> View archives and manage your subscription(s) at
>>>>
>>> http://peach.ease.lsoft.com/archives
>>>
>>>>
>>>>
>>> ===================================
>>> View archives and manage your subscription(s) at
>>> http://peach.ease.lsoft.com/archives
>>>
>>>
>> ===================================
>> View archives and manage your subscription(s) at
>> http://peach.ease.lsoft.com/archives
>>
>
> ===================================
> View archives and manage your subscription(s) at
> http://peach.ease.lsoft.com/archives
>

===================================
View archives and manage your subscription(s) at http://peach.ease.lsoft.com/archives