On Thu, 31 Jan 2019 at 17:15, Guillermo Polito <guillermopolito@gmail.com> wrote:

Hi Thomas,

Check the pragma��debuggerCompleteToSender and its users. I think it's related to that.
BenComan probably can give a better insight since (if I recall correctly) he introduced it?

Thats right.�� Use "Tools > Finder > Source" to search for... debuggerCompleteToSender��

ignoring where its a pragma.

That leaves...

�� Process>>complete:

�� Process>>stepToHome:

�� GTGenericStackDebugger>>filteredStack

�� GTGenericStackDebugger>>setDebuggerToFirstNonFilteredContext

��

The first was me. Not sure about the second. The latter two were someone else.

Guille

On Wed, Jan 30, 2019 at 5:44 PM Thomas Dupriez <tdupriez@ens-paris-saclay.fr> wrote:

Hello,

Does someone know which is the code responsible for the debugger opening on the halt of the user instead of Object>>halt (see picture)? It's neat but it's not the regular behaviour so I'm guessing it is handled somewhere specifically for halt messages?

Its not just for halt, but for any method that you don't want the debugger opening in the method sending #signal.

Is your snapshot based on modified code? In Pharo-7.0.1+build.146.sha.c48755acc5f489337be0e7abc882c15b1707b254 (32 Bit)

in Playground I evaluate...

�� 1 + 1.

�� self halt.

�� 2 + 2.

and my top line is the same as your second line.

cheers -ben

��