Going off topic, I gotta also say, coming from the php(ughh)/linux(getting there, thanks Ubuntu) world and having worked in .NET for the last 2 year, has made me more rational in my decision making process. .NET is quite sexy, for what it is. So is the DLR. So is Silverlight. The fact that we have Mono & Moonlight (to keep us ‘hippies’ happy), plus the weight that Microsoft is putting behind Silverlight, it really looks like it is shaping up to give Flash/Java/* a real run for their money. And, seriously, Ruby and Python in the browser? Hats off fellas, in this case anyhow. Now if only they would open source the .NET Javascript compiler/interpreter. It is slightly strange that they havent, but then again this smacks of latter-day decision making approaches from within the Redmond camp: Hoard Hoard Hoard. Combined with their fumbling attempts at coming to terms with an Open Source world. Hmm, perhaps this is the makings of a new post.

* Hard Mathematics (I didnt really like school, & it is just too terse)

* Unit Testing (Sounds like more work, and my boss is always asking me when the next features will be ready). My poor ol’ Tests directory on the project I wrote in Castle Monorail looks a tad bare, to say the least. (Yikes, who will employ me now?)

* Unicode. *shudder*.

OK, so moving through the inevitable effect *that* word has on most programmers. Ummm, WTF is Unicode, and how do I use it? Actually I understand conceptually what Unicode is (yes, if a language has more than 256 characters, it can’t be stored in a byte, mmmkay?) no problem there, its just when we get to BOMs, code pages and encodings, that the mental wheels start to fall off.

So this is a nice article, brushing over the bare necessities of Unicode. I have the feeling that it wont get me far, but it has got me thinking about the possibilities & ramifications of Unicode on the system I am working on. We have a client or two in India, and I have a feeling that multi-lingual is going to be a selling point in the nearish future.

This is going to come up again, so I will keep on the lookout for more infos. If anyone has any suggested resources (Im probably off to wikipedia now, so dont bother with that) let me know.

Code:

http://code.google.com/p/migratordotnet/

(No idea what is with the /p/, but i got another project under p, strange)

Mailing list:

http://groups.google.com/group/migratordotnet-devel

I got a couple of patches within a couple of days, and … *gulp* I aint got much time to devote to this (unless it’s work related).

Goody gum drops.

Check out their latest video. It’s nice to see that they don’t just write off everything except for Rails. Funny shit.

I have a branch with the fixes I have from my local repository (the pain of patching caused me to propose this to Marc):

http://code.macournoyer.com/svn/migrator/branches/refactoring

This will be a drop-in replacement for the current trunk. But I have not ported the code for Postgres or SqlServer. Postgres will probably come soon. SqLite if I have time.

Features in the branch:

Major:

Refactorings to make it easier to maintain the different DB providers. Namely each provider implements an interface, that maps to the different types, so adding new types should be easier (i.e. break the build until the interface is satisfied).

New ColumnOptions enumerations. For Indexing, Unsigned, Foreign Keys, etc.

ForeignKeyConstraint enumeration for the expected action when enforcing constraints on foreign keys. This has only been implement on mysql, so the options may not make sense for other DBs. Not sure.

Some Foreign Key generation methods, that generate the FK_name automatically. This will be worked on a bit more, as taking down the FK currently requires you to guess the FK name still.

The Migration class accepts the args from MigratorConsole. I needed this in my project, related to the next change.

The Migration class also has a virtual InitialiseOnce method, that can be overridden in a base class of your migrations, so that things like setting up Castle.ActiveRecord can be performed (Setting up ActiveRecord needs to be done in the currently running assembly to work sanely). This is used for setting up inital data (i.e. groups etc.), in a DB agnostic way. Just like the migrations do.

OK, so its quite a few changes, but patching is PITA, & I left it for too long.

Please checkout, see if your code still works. I’d be interested to know how many others use Migration.NET

Oh, and if anyone want to contact me (whups!) code+nick.hems at g-to-the-mail.com. (thats gmail guys).

The first is the fact that I have alot of dependencies in javascript files, the front page of the application (& probably where the user will spend a fair amount of time) in particular depends upon Prototype, Scriptaculous, Plotkit & Mochikit and about 20 other random small javascript files, that may or may not need to be included. At the moment my BaseController exposes functions to bring in those javascript files, and code in my layout #includes the relevant nvelocity files. This is ascetically ugly as I am calling these in my controller methods (although it could be done in the views via $controller.IncludeSomeFile). Basically I Got It Working, knowing full well that it was going to be a thorn in my side, ascetically & performance wise. So I want something akin to the Helper attributes that you can decorate your controllers (and perhaps methods) with.

The current Helper system in MonoRail works fine with the scriptaculous & Prototype Helpers at the moment, but the current situation doesnt help with javascript dependencies, or caching/bundling the javascript. Ive learnt to live with manually dealing with dependencies, so that’s not a huge problem per se (I have been developing in .NET for a couple of years now , but I have a sneaking suspicion that the 40+ javascript files being pulled in is going to exacerbate performance problems, both on the client and server.

Starting afresh on an related project, allowed me to look afresh at the MonoRail offerings, and I of course first went to the ajax related stuff. Nice to see a BehaviourHelper, but my last project used event-selectors as I liked how it works, and hinges on Prototypes $$() support, with the peformance gains and css-selector support that engenders. I also don’t like ‘meta-programming’ as it pertains to the web, as it reminds me too much of the ol’ PHP days:

$BehaviourHelper.Register("someId", "function(el){doSomething(*);}")

my javascript function often ended up with more that just a one liner, very ugly in quoted code within a nvelocity file (no syntax!). So I first baked up a solution for EventSelectors, the patch is here. Read the patch comments for what it does, & how it builds a javascript file that contains all the relevant event selectors for that area/controller/view. I really do like convention over configuration. I hope it doesnt get lost in the castle-devel deluge…

So that’s event selectors sorted, next up I need to look at:

• Bringing in ‘standard’ javascript libraries, which may also reference css files.
• Handling dependency trees (i.e scriptaculous upon prototype)
• Bringing in application javascript libraries, using a convention based approach, so your javascript application code is packaged cleanly in a folder structure.
• Handling caching of the dependency trees, and outputting concatenated/compressed javascript, just like RoR!

I’ll be putting up a class designer of my initial class design in my next post.

Shit, cool. Granted this is very doable (now that I have thought about it) using linux: livecd + install program running in background. But it is very cool nonetheless.

I also setup this wordpress blog (I could import old stuff from blogger), and took this screenshot. wow, nice. OK, so the install took 30 minutes, my personal best, although I must admit, I didnt really do anything. Time to go get a bottle of wine while I reboot the computer.

OK, glass of vino in hand. Time to setup Beryl, and the wiimote I got today.back soon.