Re: [RFA/Ada] Implement Ada tasking support (take 2)

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Joel" == Joel Brobecker <brobecker@adacore.com> writes:

Joel>       However, I think we ought to be consistent about this - I tried
Joel>       searching for other language-specific commands, and indeed,
Joel>       objc-lang defines "info selectors", "info classes", etc.

For "selectors", I don't know; but "info classes" seems like an odd
name for an ObjC-specific command.  Lots of languages have classes.
The language-specific "catch" subcommands are also a bit odd, to me.

Joel>       Tom suggested a different syntax, which is based on using
Joel>       convenience functions in the condition field.
Joel>           break LINESPEC if $current_ada_task() == 5
Joel>       It's an interesting suggestion, and I'm not opposed. I'm just
Joel>       slightly concerned that the performance cost of evaluating
Joel>       a condition expression is higher than just a pure check
Joel>       directly coded in GDB.  It's also different from what is done
Joel>       for threads.

I doubt the overhead is very high, but we can discuss that later if
you want.  It isn't relevant to the current patch :-)

I thought I heard an argument to the effect that thread-specificity
could be more efficiently implemented at a low level, at least in
theory, on some platforms.  (I don't know.  And that is mighty vague.)
If this idea applied to task-specific breakpoints then that would be
an argument in favor of putting it in the core, at least IMO.

Tom