setLibraryDebuggable

[Devon Carew]

What does this default to for libraries? All are debuggable by default?

--

Devon Carew

Software Engineer

Google, Inc.

[Todd Turnidge]

All libraries are debuggable by default.  We should probably document this.

--
You received this message because you are subscribed to the Google Groups "Dart VM Observatory Discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to observatory-dis...@dartlang.org.

[Devon Carew]

I'm happy to set the configuration I want on connection. It would be good to document that clients shouldn't rely on any specific defaults.

FWIW, the settings I'll configure will be:

• all libraries debuggable
• break on uncaught exceptions

and will follow that up at some point with a way for the user to configure it.

I don't believe there is a way for clients to query the current pause mode; perhaps that call has not been documented? I do see the recent 'setExceptionPauseMode' call.

On Fri, Oct 16, 2015 at 7:14 AM, John McCutchan <jo...@johnmccutchan.com> wrote:

I'd rather not document any defaults because I worry about people writing service clients that have assumptions like this baked in. For example, what if the program was launched with some undocumented command line flag that flipped the default?

In other words, it's unsafe for the clients to assume the default because the default could be overridden by a command line flag (or have already been changed by some other service client).

It's trivial for a service client to determine whether a library is debuggable (similarly what an isolate's exception pause mode is- see other thread). Clients should be written in such a way that they query the state of the world rather than assuming any defaults. I think that's a more robust pattern to follow.

John

[Devon Carew]

It's not critical - I plan on setting each isolate to break on exceptions as I get runnable isolate events - but I don't see the pause mode in the spec for class Isolate.

On Fri, Oct 16, 2015 at 9:10 AM, John McCutchan <jo...@johnmccutchan.com> wrote:

The exception pause mode is returned with the Isolate.

John

[John McCutchan]

This is landed now. 

John

On Fri, Oct 16, 2015 at 12:24 PM John McCutchan <jo...@johnmccutchan.com> wrote:

Yep- I forgot to add it to the spec:

https://codereview.chromium.org/1408933004

[John McCutchan]

The exception pause mode is returned with the Isolate.

John

On Fri, Oct 16, 2015, 8:33 AM Devon Carew <devon...@google.com> wrote:

[John McCutchan]

Yep- I forgot to add it to the spec:

https://codereview.chromium.org/1408933004

On Fri, Oct 16, 2015 at 12:20 PM Devon Carew <devon...@google.com> wrote:

[John McCutchan]

I'd rather not document any defaults because I worry about people writing service clients that have assumptions like this baked in. For example, what if the program was launched with some undocumented command line flag that flipped the default?

In other words, it's unsafe for the clients to assume the default because the default could be overridden by a command line flag (or have already been changed by some other service client).

It's trivial for a service client to determine whether a library is debuggable (similarly what an isolate's exception pause mode is- see other thread). Clients should be written in such a way that they query the state of the world rather than assuming any defaults. I think that's a more robust pattern to follow.

John