Mailing-List: contact derby-user-help@db.apache.org; run by ezmlm
Precedence: bulk
Reply-To: "Derby Discussion" <derby-user@db.apache.org>
Received-SPF: pass (hermes.apache.org: local policy)
Mime-Version: 1.0 (Apple Message framework v619)
Content-Transfer-Encoding: 7bit
Message-Id: <E97D19C0-19E6-11D9-A96B-000D93ADDBD4@serv.net>
Content-Type: text/plain; charset=US-ASCII; format=flowed
To: derby-user@db.apache.org
From: Joseph Grace <occam@serv.net>
Subject: FAQ language (was: Unable to run 10.0.2 GA on Mac OS X 10.3.5/VM
 1.4.2_05)
Date: Sat, 9 Oct 2004 04:32:33 -0700

Dear Derby-user:

First of all, I'd like to thank Jean and all the IBM folk for making 
this transition smoothly and proactively to get things done, like the 
new FAQ.  I would like to suggest an adjustment to the language of one 
FAQ entry though.

On Fri, 08 Oct 2004 10:26:02 -0700, Jean T. Anderson 
<jta@bristowhill.com> wrote:
> Incidentally, this problem is now on the FAQ page

+1

>  (and Suresh's summary
> of the problem was so helpful I used it almost 'as is'):
>
> http://incubator.apache.org/derby/faq.html#createdb_OS_X

-1 since that FAQ entry says:

> 2.1. Why can't I create a new database on Mac OS X?
>
>  This is a JVM problem on Mac [...]

Jean, I suggest changing the language of this FAQ entry to be fairer to 
the JVM team at Apple.  AFAIK, noone has resolved the issue to a bug in 
the Mac JVM.  At best, the rws/rwd problem is a

     "Derby issue with Mac" OS X

+1

not necessarily a Derby "problem on Mac" OS X.  It's a subtle but I 
believe important distinction to avoid false accusations, especially 
with those whose help we may end up needing and hoping to enlist for 
insight and solutions to gray area issues such as this one.

Someone mentioned that they filed a report to the Mac OSX team which is 
great, so at least Derby is (fingers crossed) already on the Apple Mac 
OS X team radar.  I don't think a copy of the issue report was posted 
to either list, so I don't believe I've seen the submission but do hope 
the submission was polite and simply flagged the issue without 
necessarily laying blame.

For a full technical explanation of the issue, check my posts to the 
archives of derby-dev describing the rws/rwd issue when running Derby 
on Mac OS X.  The Jira log for this bug has a few misleading (implying 
blame) entries which do not follow from my analysis of the situation.  
I chose not to clarify the issue further since it seemed picky at the 
time since I did not think the language in the Jira was important 
enough to address (except by example) since Jira does not seem all that 
public, and the project is just getting started.  Perhaps that was a 
mistake.  I do think the language for the FAQ is important however, 
since the FAQ is a genuinely public document.

> The workaround is to set  derby.storage.fileSyncTransactionLog=true  
> to use file sync instead of write sync for log writes.

+1

As for this actual workaround, I believe that's fine.  As I understand 
it, it sacrifices performance to avoid the gray area of rws/rwd 
implementations, and in so doing avoids code changes (as in my original 
workaround/potential fix).  Definitely a good getting-started 
workaround even if not necessarily the appropriate long-term solution.

I have some ideas for further examining this issue but just haven't had 
the time to do the exploratory work.  AFAIK, my invitations for more 
information came up empty, so I'll repeat the request.

If anyone has more light to shed on the (general JVM or specifically 
Mac OS X) rws/rwd issue, please do so.

Thanks,

= Joe =