jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jukka Zitting ...@yukatan.fi>
Subject Re: JCR-RMI diffs
Date Sun, 08 May 2005 13:40:54 GMT

Ludovic Maitre wrote:
> I have do some modifications to JCR RMI so it is compatible with the 
> last SVN release of Jackrabbit, i hope.

Very nice, thanks! However, I'm sorry for being too quiet about working 
with Felix Meschberger to do the same upgrade. :-( I'm expecting a final 
patch from Felix within a few days.

I hope you are not taken back about this double work. I reviewed your 
patch and it is fine except for the following issues.

> The changes are:
> - add a dependency to jackrabbit to use his valueholders class
> (LongValue...)

I'm not too happy about introducing a direct Jackrabbit dependency into 
JCR-RMI. The idea of JCR-RMI is to be a general RMI binding to various 
JCR repositories. A direct Jackrabbit dependency breaks this generality.

In this case the problem is not as big as it seems, as the dependency 
only covers the Value implementation classes. It is possible to simply 
copy those classes with minor modifications into JCR-RMI to avoid the 
full dependency. (This is actually what Felix was doing.)

Better yet, we've been discussing with Felix about the possibility of 
making this part of JCR-RMI into a general purpose Value implementation 
that could be placed in the JCR-EXT package. More on this in a few days.

> - add the getValueFactory() to the remote and local sessions. I'm really 
> not sure if we should permit to get the valuefactory, but i have done 
> this so this compile.

I don't think the ValueFactory interface needs to be exported over RMI. 
Just make ClientSession maintain it's own ValueFactory implementation.

> - add an ignore rule for this new method into the test because i don't 
> know what is appropriate to use it.

That's fine.

> So if somebody will check if this modifications are ok, they can be 
> committed.

Based on the unnecessary Jackrabbit dependency and the upcoming patch 
from Felix, I must unfortunately reject your patch. I hope you are OK 
with this and understand the reasoning.

> I must precise that i have never use the jcr-rmi ext before so i perhaps 
> that i have modified this in an inapproriate manner.

I appreciate your effort and initiative! Unfortunately it got somewhat 
wasted because of little communication on my part. I hope seeing more 
contributions from you!


Jukka Zitting

View raw message