Re: What does maxquerylen in v$undostat really mean?

From: Howard J. Rogers <hjr_at_dizwell.com>
Date: Sun, 12 Dec 2004 18:39:18 +1100
Message-ID: <41bbf5a6$0$1082$afc38c87@news.optusnet.com.au>

HansF wrote:
> Howard J. Rogers wrote:
>

[snip]

>
>
> Based on my experience, I find Oracle tends to retain definitions for a
> term - if they want to change the definition, they tend to change the name.

"Experience". "Tend". These are generalities. Documentation should provide specific answers to specific questions. Your comment misses the point that, without specific, accurate documentation, we cannot know *for certain* whether your experience is playing us false this time; or whether the perceived past "tendency" not to change definitions is in force this time round. We cannot know. Therefore we are guessing and relying on our subjective experiences and perception of tendencies... and that is precisely what documentation is there to avoid.

> If I find something in a specific version, and I can't see a reference
> in the documentation, I'll use other tactics to build a temporary image
> as a basis for investigation - and then verify using other testing
> techniques (including asking here and looking in metalink). I see
> nothing dangerous in that.
>

>>
>>> One public place to look for information known to be missing, or 
>>> incorrect is in the next version of the docco.
>>
>>
>>
>> I really don't know whether I'd go that far. Certainly, it would 
>> provide a probablistic answer -and maybe, the probability might be 
>> quite high. But it is never simply 'the answer' -which requires Oracle 
>> Corp. gets its act together and fixes its documentation when the error 
>> is pointed out.
>>

>
> I did not state that this would yield *the* answer. I stated this was
> an investigative technique.

I didn't claim that you stated it would yield the answer. I drew the obvious and logical inference from the words you *did* use. Namely: one investigates something to *find an answer*. You don't investigate to find the wrong answer, at least, do you? Therefore, investigation is for a purpose, and to a point. The mere fact you mentioned "investigation techniques" therefore implies that reading the wrong version of the documentation can help you find an answer.

I disagree with that. Whatever your personal experience, it is I think bad practice to *assume* that because something is mentioned in version 10's documentation, it must also have been that way in version 9. You might remember, for example, the glitch that affected tkprof when they forgot to ship a 9i-specific version with 9i Release 1: if you ran 8i's version against a 9i trace file, your elapsed times (ironically enough) were out by a factor of 10.

So yeah. Your investigation technique can probabilistically help find an answer/the answer. But there is a probability, too, going across versions especially, that the answer will be wrong, and hence not a useful answer at all.

And what there is for you to find exceptional about that, I cannot fathom.

>>> After all, Howard had already stated it's missing in the 9i docco ... 
>>> and we already know that Oracle will not generally release updates 
>>> (ie. patches) to the software on OTN, instead waiting on the next 
>>> public release.  Can't see why we'd expect anything different from 
>>> the docco which, like software, is subject to the possibility of bugs.
>>
>>
>>
>> It seems to me that the software issue is different from the 
>> documentation one, as evidenced by the fact that OTN itself talks 
>> about 'trial licenses' and 'evaluation purposes' and so on for the 
>> free software downloads, but makes no such qualification regarding the 
>> contents of tahiti.oracle.com.
>>

>
> Quoting from
> http://download-west.oracle.com/docs/cd/B10501_01/server.920/a96521/title.htm
>
>
> "The information contained in this document is subject to change without
> notice. If you find any problems in the documentation, please report
> them to us in writing. Oracle Corporation does not warrant that this
> document is error-free."
>
> Is that not sufficient qualification?
>
> LWIY/Hans

I don't think we need start WWIII over this. No, it's not a sufficient qualification. It's an invitation from the Corporation to get a documentation error corrected (which is most welcome, of course). But that is rather different from your original "they don't update the free software downloads as patches are released, so why should we expect them to do anything different with the documentation". The quotation you cite itself indicates that we very much *should* expect something different.

Regards
HJR Received on Sun Dec 12 2004 - 01:39:18 CST