I read a bit about IoC, and while I think that it's a useful pattern, I've not used it in my applications. Mainly for lack of need and reluctance to complex my life even further :-). Not to mention that I dislike learning another XML Dialect.

After reading this article, it seems that IoC can be much better than the current popular implementations.

The gist of the article is using an interface that encapsulate all the external needs for the class, which allows for greater flexibility and testability. I'll probably use it on the next application that I'll create.

I seems to have misplaced my weekend, since it just vanished without a trace. All I did was catch up on my reading and then handle some personal matters and then it whisked away.

I got a note that I might meet it in a few days, but only if I behave. Do you think that I need to post a carton milk shot to find it?

On other news, I found  The Golem's Eye in a book store yesterday, I very much enjoyed the  The Amulet of Samarkand, so I look forward to read it. 

No, I'm not going to complain about something here.

I returned from a trip the the lowest place on the world, 408 meters below the sea level. The Dead Sea.

The view was fantastic, and convinced me that I really do need a digital camera :-)

Hebrew speakers will cring at the translation, but this is what passed through my head there:

All the new code is in the repository now, you can check it out using:  svn checkout svn://svn.berlios.de/nqa

Planned features are:

• GUI for creating hbm files.
• GUI for creating cfg.xml files.
• Generating code and assemblies from hbm files using the current tools.
• Anything else you would want, just tell me :-)

The HaloScan.com team has fixed the vulnerability in their services :-)

And in a very good time, too.

New release, new features :D

• Ability to see NHibernate logs, and filter them by logger, severity, text and using a regular expression. [This is implemented as a user control, so it may be of use in some other projects as well]
• Fixed the performance counters to disregard the time it takes to display in the UI. The results from my informal testing are very good for NHibernate (again, do your own testing, YMMV)
• Added the ability to select projects (which I for some reason forgot to add :-) )

That is all for now, feature requests are always welcome.

Get it at: http://developer.berlios.de/projects/nqa/ 

The reason that I choose BerliOS is that they provided me with subversion access. But, what I didn't counted on was that I would usually develop against my main repository at home.

Now I've a situation where I've two repositories, of the same data, that I want to keep in sync. One way to do this would be to just dump & load the data, but that seems... crude.

Anyone can recommend another way?

Pragmatic Version Control Using Subversion  is a book you need to read if you work with Subversion. And I feel confident saying this even though I've yet to read the book.

I've read Pragmatic Version Control Using CVS and that was good, even though I don't use CVS.

The pargmatic bookshelf have not let me down yet, they are interesting, full of pragmatic :D information and useful.

Now if I could find out how to work with two repositories in parallel, that would be great :-)

Just something interesting that I noticed when fixing some things in NQA.

I've added performance measurements and I got strange results, I was getting bare database access times that were longer then the NHibernate times.

That didn't feel right, I wasn't doing anything very serious, but I knew that NHibernate must be doing something, and that meant that there had to be some overhead. Certainly not the other way around.

When I took a look inside I discovered that I was timing UI updating as well. [Yes, this is bad, but that was the fastest way to get the code done at the time.] I refactored it a bit, so now it pulls all the data from the database first and measure only that.

The interesting thing is that the times went from ~900ms to ~250ms. Perhaps I'm very inefficent in my UI (likely, even) but my conclusion¹ from this data is that the UI is very slow. Has it been your experiance as well? {I was updating a ListView, btw}

¹This is based on one case only, so don't take it too seriously. It may be that I do inefficent things in the UI.

"Never violate the Prime Directory! C:\" -- Tech Support Slogan

"Someday we'll look back on all this and plow into a parked car." -- Bumper Sticker

"Strange is our Situation Here Upon Earth." -- Albert Einstein

I just got the standard referrer email from my blog on my latest post. Only this one wasn't the usual one. I usually check the pages that link to my blog if I can, and so I followed this one, to find myself in Haloscan's page.

The problem that I have with is simple, actually. The url that I got had the normal format of http get parameters, and allowed me unlimited access to the account. Including editing/deleting both comments and trackbacks.

What happened is that I trackbacked the story, and the author apperantly followed it from his management page to my blog, which caused my blog to send me a referrer email with the referring adress.

The big problem here is that just the URL is enough to warrant access to the account.
At a minimum, they need to use cookies or POST variables.

I informed both the author and HaloScan.com, we'll see what comes out of it.

This is the first security related bug that I discovered. :-)

[Update: It seems that you can mitigate this if you logout of your session]