Recursive Runner class to get report on dir size

[aratikakadiya]

No description provided.

[pdowler]

Just review of main logic so far

[pdowler]

Is this needed and used now output is written to a storage URL?

[pdowler]

It is best to do this at the start of performAction and not in setJob; this method should probably not throw

I realise the other impls do some work in setJob and I will fix that and not override this method.

reason: exception throwing and handling needs to be in JobRunner.run(); the base class does all that and calls performAction() and handles all failures from it

[pdowler]

I think I grok how the above expression will work. upload.run() will make the call to putURL, check errors, and then open the output and call OutputStreamWrapper.write(OutputStream). The thing I don't like about the above expression style is it never actually says it is implementing that write method and the stream -> in the HttpUpload ctor call is very subtle. aside: the out variable here is actually a member of the outer class; I usually try to make assignment to class variables stand out with this.out = ....

This version has made the complication of trying to do a lazy init and use OutputStreamWrapper clear. I think there is a way to do it that would work in the typical case (depth=0 and large allocation)...

I think this can be made more clear with a concrete nested class eg:

private class DynamicReportOutput implements OutputStreamWrapper {
    
    private PrintWriter output;

    DynamicReportOutput(ContainerNode cn, Subject caller, int depth) {
        // TODO: scan up to first output here??
    }

    public void write(OutputStream ostream) {
        this.output = new PrintWriter(new OutputStreamWriter(ostream));
        ...
    }
}

I would probably move as much state as possible inside this nested class, and maybe even the accumulate and output methods from the parent (eg the Writer declared above).

I think we could make this do a lazy init with a method in the nested class that scans/accumulates up to the point where it has to do output and then stops. At that point the HttpUpload is created and run(). I think doing that in the ctor would make usage very straightforward... this is typical of a dynamic Iterator (they advance to the first item in the ctor).