How LINQPad Works
LINQPad is a client/server application with a twist. Whereas with most client/server
applications, there are many clients and one server, with LINQPad there is one client
and many servers!
The user interface is the client, as you might expect. For each query, LINQPad creates
a separate server, which a class that runs in its own process and executes
the query in isolation. This isolation prevents queries from interfering with each
other (or the UI) and allows LINQPad to safely cancel a query without polluting other application domains.
In the past, LINQPad used in-memory Remoting channels to communicate with the server processes. Now it
uses a custom-written communications layer that runs atop Windows shared memory (yes, there are plenty of pointers!)
This provides a faster, more reliable, and fully asynchronous communications layer.
LINQPad 5 compiles your queries using the Microsoft Roslyn libraries (in the past it used .NET's CSharpCodeProvider and VBCodeProvider).
Because C# and VB are statically typed, any database objects that you reference
need the backing of a typed DataContext. For performance, LINQPad builds typed DataContexts
on the fly using Reflection.Emit rather than generating and compiling source code.
It uses LINQ to SQL rather than Entity Framework because LINQ to SQL is lighter and faster in building the metamodel when instantiated.
Reflection.Emit must instantiate types in memory whose assemblies cannot
be unloaded. For this reason, LINQPad's emission engine itself runs in a separate
application domain, which is regularly refreshed to avoid an ever-increasing memory
Because so much work goes on the background (querying database schemas, emitting
typed DataContexts, compiling and then executing queries), every potentially
of LINQPad must operate asynchronously to maintain a responsive user interface.
The interaction between the many threads that perform this work and the main thread
driving the UI (with endpoints such as updating the repository tree view and autocompletion
service) was one of the hardest things to get right!
LINQPad's Dump command feeds the output into an XHTML stream which it
displays using an embedded web browser (you can see this by right-clicking a query
result and choosing 'View Source'. The transformation into XHTML is done
entirely using LINQ to XML, as one LINQ query! The deferred expansion of results and
THis means that most of the XHTML is typically prepopulated after a query finishes
executing. The lambda window populates using a custom expression tree visitor
(simply calling ToString on an expression tree is no good because it puts the entire
output on one line).
LINQPad uses a number of third-party components within the UI. The query editor
uses Actipro's SyntaxEditor control
(a very polished product);
the "IntelliSense" features rely on a combination of libraries from Actipro
Software and ICSharpCode (from the SharpDevelop project). Here's how it all fits together:
Features such as syntax highlighting, red squiggly underlines and autocompletion
require that you lexically and semantically parse the source code. Lexical parsing
means reading the raw text stream and breaking it up into a stream of tokens; a
semantic parser then reads those tokens and figures out what they mean in context,
emitting a code DOM (called an Abstract Syntax Tree). The final step is to resolve
the nodes in the AST into .NET types and members (binding).
LINQPad 5 uses the Microsoft 'Roslyn' libraries for parsing, binding (and compiling).
Prior to this, the heavy lifting was done by NRefactory and SharpDevelop.Dom, written by
Daniel Grunwald, Andrea Paatz, and Mike Krüger (as part of the SharpDevelop project).
These are fantastic libraries: writing a fully-featured lexical and semantic
language parser in itself is no easy task; writing a well-structured type resolution
DOM that can correctly handle the nuances of C#'s type inference is a major achievement.
The debugger, too, uses a library that builds on NRefactory (for the watch window expression evaluator).
The low-level work is done by an adapted version of Microsoft's Managed Debugger sample. Two additional
layers sit on top of this, which are part of LINQPad. The first provides a lazy, asynchronous, cancellable, mostly immutable layer
which can self-heal references that have been moved by the GC. The second is a WPF View-Model, which synchronizes
with the layer below to provide bindable objects to display in the list views.
Because LINQPad ships as a single click-and-run executable, all the libraries are
included as embedded resources. (Look at the last example
on this page to see how). A bonus of this approach is that you can compress the
libraries at the same time, reducing the executable size.
LINQPad automatically patches itself by downloading updates into its Application
Data folder. It then checks that the new assembly has the same strong name, and if so,
forwards to that executable, which then writes itself back to the original file.
This may fail (depending on the permissions of the folder to which LINQPad was downloaded) so LINQPad always checks
whether there's a later signed version of itself
in its Application Data folder when started.