Re: [linuxtools-dev] [TMF] New Features in TMF about filters expressions

[Florian Wininger <Florian.Wininger@xxxxxxxxxx>]

Hi Matthew,


On 13-03-12 11:31 AM, Matthew Khouzam wrote:

Hey Florian,

Welcome to the cyberspace bulletin board of linux tools. Great to see your feedback! So first, could you differentiate string from identifiers? Is RUNNING a value or a variable? I'm asking because if it's ambiguous.

>If you dont have relative paths, or a special syntax for relative paths, RUNNING would be a value.

Yes actually, RUNNING is a value.
For relative paths, a special syntax would be add with the prefix ./

Can we have "/jazz/blues/artist == ${/artist}" ?

Yes, we can have this and this syntax is equivalent to :

/jazz/blue/artist == /artist

For your syntax ${} operator look for the value of /artist and replace this value before interpret the comparison.
/jazz/blues/artist == ${/artist}
=> /jazz/blues/artist == artistname
(artistname is a value of /artist)

Just out of curiosity, why not substring searches. Here's a real world example.

You have a multi-threaded game you're tracing, let's call it "Doom"
Now it registers thread names as "Doom:PIDNum" or "Doom:audio" "Doom:video" "Doom:AI" "Doom:physics[num]"
I think people would want- No- DEMAND! ;) that you have
/processes/*/name.contains("Doom")
                  ^---- Wildcard

Substring search and wildcard are two feature interesting to add. I thinks about that.

Also, the Java .equals may be less ambiguous than the C == for string comparisons.

Why not.

I was thinking over lunch, why would we need assignation in the specification of the filter? 

Assignation is not necessary in the specification of the filter unless it's to implement local variable in the filter.
However assignations are here to doing the samething that would be done in the Java state provider and make a build-your-own-state-provider to replace the CtfKernelStateInput.java



Thank you for your comments,
Florian