RFC0011-Dot-Referencing-With-Static-Integers.md feedback about powershell-rfc

[KirkMunro]

Issue for comments to: https://github.com/PowerShell/PowerShell-RFC/blob/master/1-Draft/RFC0011-Dot-Referencing-With-Static-Integers.md.

[lzybkr]

Two comments:

First, the RFC doesn't mention the most likely break - running a native command like 42.exe.

Today, if you have a native command that is a decimal number, it can't be invoked with just the number, so using the extension is a natural solution (probably more natural than & 42.

Second, we need to be sure not to break parsing command arguments, e.g.

echo 1. # Note, no property name

I can easily see this in real world code and parsing as an expression, we'd report a parse error. And no property name shouldn't be a special case, echo 4.thing should print 4.thing, not the empty string (assuming no property thing exists.

[KirkMunro]

Thanks for the comments @lzybkr.

At the end of the Specification section, the single con that is listed is about that break. I didn't specifically call out invocation of a .exe whose name is an integer though.

On your second point, in the referenced implementation for this (which is linked at the bottom of the RFC), I get the results you expect, which actually both surprises and disappoints me a little. The parser doesn't seem to resolve properties or methods on numeric literals when parsing command parameters in PS5 or PS6. String literals, yes, but numeric literals, no. For example in both versions, echo 'abc'.GetHashCode() shows me the hashcode for that string, but echo 2l.GetHashCode() returns a parser error. On the plus side for what you mentioned above, how it works for parameters today doesn't change. On the downside, it means to dot reference any member (property or method) on a numeric literal in a command invocation, you need brackets surrounding the entire thing. At least that would be consistent regardless of the format of numeric literal you use though if this RFC gets approved.

Here is some output from my PS Core dev environment with the parser changes I implemented to try this RFC out:

PS C:\> $PSVersionTable.PSVersion

Major Minor Patch Label
----- ----- ----- -----
    6     0     0 alpha


PS C:\> Update-TypeData -TypeName int -MemberType ScriptProperty -MemberName Hours -Value {New-TimeSpan -Hours $this}
PS C:\> 2.Hours.ToString()
02:00:00
PS C:\> echo 2.Hours
2.Hours
PS C:\> echo 1.
1.
PS C:\> echo 4.Thing
4.Thing

[KirkMunro]

@lzybkr: This is past the 30 day mark. What's next? Discuss internally?

[SteveL-MSFT]

@PowerShell/powershell-committee to discuss