db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Tiago Espinha <tiago.de...@yahoo.co.uk>
Subject Re: Database name length
Date Tue, 14 Sep 2010 13:35:43 GMT
Hi Rick,

Unfortunately I only read your mail after I sent mine but I still believe that 
whether we violate DRDA or not, these should be kept as separate issues. One: 
implementing the UTF-8 support and the other, extending the allowed length. This 
allows me to go ahead and commit my work on the UTF-8 support soon and we can 
have it on time for 10.7.

Does anyone disagree with this?

Tiago


----- Original Message ----
From: Rick Hillegas <rick.hillegas@oracle.com>
To: derby-dev@db.apache.org
Sent: Tue, 14 September, 2010 13:50:35
Subject: Re: Database name length

Tiago Espinha wrote:
> Indeed, the length would still take two bytes and with these two bytes we'd be 
>able to take it up to 65535 in length.
> 
> After talking with Kathey on IRC, I do agree that we might be deviating from 
>the standard with this. Can we still claim to be standard-compliant if we make 
>an exception for this. Can we pin this as an extension to DRDA?
> 
> In truth, making this change would be as easy as changing a codepoint but I 
>think the issue here is standard compliance.
>  
Hi Tiago,

DRDA is not one of Derby's governing standards. It is not mentioned in Derby's 
charter and the community has never voted to require DRDA compliance. We are 
free not just to extend DRDA but to flatly violate it if we need to.  We use 
DRDA because:

1) It is a public protocol which has been designed carefully.

2) It actually works, as demonstrated by a couple implementations.

3) When we need new network functionality, we often find supporting protocol 
already defined in this public, thoughtful, proven standard.

We do not make a general guarantee that any DRDA-speaking driver can communicate 
with the Derby server. I am fairly confident that the Derby client is the only 
DRDA-speaking driver that works with our server. It is certainly the only driver 
we bother to test. A driver written for DB2 does communicate with early versions 
of the Derby server. However, we have not tested that driver with recent Derby 
versions.
> Kathey suggested that we'd turn this into a new JIRA issue and that we'd try to 
>get it through opengroup to change the standard to allow larger lengths.
>  
Only one company has bothered to rent an expensive seat on the DRDA committee. 
All of the other committee members are merely advisors, who sanity-check but 
cannot initiate changes. Any change to DRDA would have to come from the 
seat-holder. His name is James Pickel and he works for IBM.

> I'm torn on this issue to be honest...
>  
I am comfortable with extensions and deviations, provided that we document them.

Hope this helps,
-Rick
> Tiago
> 
> 
> ----- Original Message ----
> From: Dag H. Wanvik <dag.wanvik@oracle.com>
> To: derby-dev@db.apache.org
> Sent: Tue, 14 September, 2010 0:02:49
> Subject: Re: Database name length
> 
> Kathey Marsden <kmarsdenderby@sbcglobal.net> writes:
> 
>  
>> Does anyone know if there are other clients besides derby client that
>> might be working and expected to continue to work against Derby?
>>    
> 
> I don't, but even if there were some, they would continue to work if
> they do indeed work with the present server, since the encoding of the
> db name length (2 bytes) would be unchanged, if I understood this
> correctly.
> 
> Dag
> 
> 
> 
>      
> 
>  


      


Mime
View raw message