Just A Programmer We're just programmers

27Mar/110

My ideas for The Atlantis Interactive Schema Engine

I first became aware of the open source Atlantis Interactive Schema Engine on Febuary 24th 2011 thanks to a retweet by Paul Randal. Its a .NET assembly which contains most of the “heavy lifting” code related to the schema operations in their various products. Matt Whitfield, the sole proprietor of Atlantis Interactive, is going to setup an account on codeplex or one of the other forges so people can post bugs, feature requests, and of course patches. Personally, I have several ideas for feature requests. However, I have no channel at the moment for these requests besides emailing Matt. While Matt has been very responsive, I wanted the conversation to be more public, I talked to Matt and he agreed it was a good idea to create put them on this blog post. Please use the comments to add your own bugs, feature requests, or patches, or comment on my feature requests.

Now, here are the feature requests.

Integration with ILSpy

ILSpy is a free and open source .NET decompiler. Development was accelerated by the fact that Redgate Reflector ceased to be free. It would be great if the library could decompile CLR assemblies to C# with this tool. It would be great to produce a diff of the decompiled source of two CLR assemblies.

Powershell related improvements

Powershell Cmdlets

I’ve written about using the schema engine with  powershell previously. Its a relatively pleasant experience, but it could be improved with powershell cmdlets. I’m pretty new to powershell, so I don’t have a clear direction on where I’d like these to go. However, I might consider taking the initiative on this one.

Shortened namespaces for the Enums

Basically, in powershell you have no using statements, and the Enums that are used as parameters are several namespaces deep. It might not make sense to change them. However, I am just throwing this out there.

More flexibility for creating SchemaReaders

I’m looking to be able to do a few things:

  • Pass an existing connection object
  • Create a pooled connection string. While that doesn’t make sense for desktop apps, it might make sense if I was using the library from a web app. Being able to sync databases from a web app would be really cool.

Get the existing connection string from  SchemaCollection

Its self explanatory.

Be able to script a schema to a Microsoft Sql Server Project

I’d like to be able to point the library at a database and have a function in SchemaCollection that will script everything to a visual studio database project. Initially it would be easiest to hard code this functionality. However, it would probably be best if it was template (t4?) based so that you could easily support other database versions.

Naturally a long term possibility would be to be able to do a schema sync between databases and sql server projects.

Support for Mono

I actually started a port of this. Its very easy to get it to compile (comment out like 3 lines), but you do lose Geospatial support. For this reason, it might make sense to have separate mono and .NET builds. In my limited time testing it I could not get my linux vm to connect to SQL server, so I don’t know if it would actually work. I want to pursue this path a bit myself, even if it just leads to me filing upstream bugs in mono.

Support for other databases

It seems like the object model is pretty well organized. That being said I realize this exercise would be far from mechanical. I don’t think cross database compares will ever be practical on a whole database level.  Below are the specific databases I’d like to see supported.

Microsoft Access

I don’t actually see much of a case for this other than “it can be done” and all windows machines have support for the JetSQL engine.

SQLite

Support for SQLite in .net is very good via System.Data.SQLite.

Postgres

Its been a while since I used NPgsql, but its pretty stable and robust. That being said, there will always be edge cases here since postgres is so customizable and extensible.

Conclusion

So thats a tall list of requests. When we have an issue tracker I will be transferring them all there. Until then, use the comments.