It seems my mind is obsessed with another crazy idea. I hope you won't mind 
if I record it here. =)

I look at recent problems mostly caused by 'too early' or 'late' execution 
of certain parts of our code. Currently, I don't have even a slightest idea 
of how Spyder initialization should work, which components should be 
initialized in which order, and - more importantly - how does it currently 
work in reality. I feel like I need a lot of time for static analysis to 
understand if it is an asynchronous run-time problem - or problem with the 
way our code is statically woven. 
(that's the reason I can't reply Carlos to commit review at 
http://code.google.com/p/spyderlib/source/detail?r=fc5092657bc3cbfd86967de55bed688892b0d0b1
 
as I have more questions pop up in my head than I can possibly squeeze into 
a small letter given the time constraints we are all in)

So, I wonder if there is time to invent something that will make analysis of 
code execution order convenient? 'Something' that is able to dispatch 
asynchronous tasks, control them and provide visual tool for run-time and 
post-mortem analysis? The use cases:

1. wait until some task finishes processing before continuing (another 
approach - wait for data in channel - Go-like)
2. profile bottlenecks and time spent waiting for components
3. debug multithread locks, inspect current Spyder state in real-time
4. visual representation of time spent in threads (bars from UML sequence 
diagram)
5. provide high-level overview of Spyder initialization sequence using 
actual real-time data (like profiling, but with high-level components or 
blocks that are convenient for the given context - i.e. trace initialization 
and wait time of various Spyder parts)

-- 
You received this message because you are subscribed to the Google Groups 
"spyder" group.
To view this discussion on the web visit 
https://groups.google.com/d/msg/spyderlib/-/6-jM_vl9SlUJ.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/spyderlib?hl=en.

Reply via email to