Raul has summarized my thinking. Debugging multithreaded code is hard,
and I think the debugger is going to have to be the main tool.
Henry Rich
On 4/11/2022 2:59 PM, Raul Miller wrote:
On Mon, Apr 11, 2022 at 2:51 PM 'Pascal Jasmin' via Beta
<[email protected]> wrote:
I'll add more simply, I don't like this display feature.
The default display should be simple and robust.
It's our natural instinct, as programmers, to want to have as much
insight as possible into the state of the things we are working on.
But, when dealing with multiple threads, that's a highly volatile
situation and anything we see will be a representation of the past.
Anyways, ... it should be possible for you to construct the inspection
mechanisms you want (with limitations imposed by the physics of time).
But the robustness of the implementation should be a higher priority.
(As an aside, as a programmer I have a *lot* of fondness for the
concept of display being equivalent to serialization. But even that
level of detail sort of falls apart for threading, because of the
transitory character of thread states.)
I hope this helps make sense of things...
--
This email has been checked for viruses by AVG.
https://www.avg.com
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
