Restrict console output for methods and types with QUIET passed

[jashook]

In addition, if JitDisasm is set skip methods which do not have the
method name. This avoids compiling all methods when there is only
one that is intended to be dumped.

[jashook]

@BruceForstall please route as necessary

[jashook]

/cc @sandreenko

[BruceForstall]

also @AndyAyersMS

[BruceForstall]

The JitDisasm changes worry me, as they're a bit of a hack: they don't parse using the same syntax as the JIT. E.g., they don't handle *. And it doesn't look like the two cases parse the same way.

The verbose changes to DriveAll seem reasonable, but I haven't been able to successfully run DriveAll locally to see what the current behavior is, so I need to investigate that first.

[jashook]

This is correct, anything with the wildcard will currently fail. Adding logic for that, would be a bit annoying but seems required.

[AndyAyersMS]

I've wanted something like this for a while, though generally I'd prefer to specify the pattern via command line args instead of env vars (which I realize is awkward because pmi's command line parsing is lame).

So maybe (in verbose) mode indicate that the results are filtered?

[jashook]

I am fine with changing this to be passed in via command line arguments. Seems to somewhat diverge from current expectations; however, seems like the correct direction.

As for the filtered indications, sounds like printing method ------ is skipped.

I've wanted something like this for a while

I found pretty quickly it is necessary for large assemblies. Between the disk usage and just the time to dasm a large assembly makes this change a requirement for viewing dasm for a particular method in most assemblies.

[jashook]

Thank you for the initial comments, I will work on the feedback. This is no where near my highest priority, so my guess is I will touch back on this in a week or two.

[BruceForstall]

fwiw, another workflow could be: superpmi collect a PMI run, then do superpmi replay with asm for a particular function or functions. This would be a one-time hit for the collection, but everything subsequent is really fast. I can give more details if interested.

[jashook]

I think the larger issue here with relying on SuperPMI is the fact that the assemblies are not snapshots. They change often and recollection is expensive and long.