Release v3.26.2.0

Significant usability improvements particularly for supporting C# 7 on Mono/Linux. As well as the features triggered by VSCode integration.

• Linux support

  • Added "/debug:pdbonly" workaround to allow generation of debug symbols on Linux+Mono. Needed as Mono "/debug+" is broken on Linux.
  • Added resolving GAC assemblies from namespaces Mono/Linux
  • Added on-fly conversion of *.pdb to *.mdb under Mono
  • Patched Microsoft.CodeDom.Providers.DotNetCompilerPlatform to allow using Roslyn on Linux. This work around allows custom compiler (csc.exe) path and handles Mono problem when interfacing Roslyn (C#7). This is a work around for https://bugzilla.xamarin.com/show_bug.cgi?id=57130
  • Moved lock objects (*.lock files) on Linux away (to cache) from the actual folders where the files to be locked are. Just to stop pouting the file system

• Deployment

  • Default CS-Script code provider renamed to CSSRoslynProvider.dll.
  • Default code provider CSSRoslynProvider.dll embeds patched Microsoft.CodeDom.Providers.DotNetCompilerPlatform.dll to allow simplified low footprint deployment.
  • Added auto-loading CSSRoslynProvider.dll if found in the script engine folder.
  • Added resolving code provider when it is specified by file name without extension

• C# 7 and general improvements

  • dbg.print extended with direct support for IDictionary.
  • Added alias to the '-provider' switch: '-pvdr'
  • Added fully named equivalent of '-s' switch: '-sample'
  • Added C# 7 specific sample with '-s:7' switch
  • Added '-tc' switch for tracing the actual compiler input for CSSRoslynProvider.dll
  • Script engine assembly (e.g. cscs.exe) has been exposed to the precompiler routines.
  • Added ScriptParser.ProcessImportedScript to be used from precompilers.

• Assorted defect fixes:

  • Issue #70: Minor issue related to platform dependent line breaks, e.g. in error messages
  • Issue #69: Basic types not available to precompilers on linux
  • Issue #68: Add option to take Main from primary script file
  • Issue #67: Same script, multiple concurrent instances, on Linux (again)