Feature #8263
open
Support discovering yield state of individual Fibers
Added by halorgium (Tim Carey-Smith) over 11 years ago.
Updated almost 3 years ago.
Description
Related to #8215, when debugging a multi-Fiber application or library it is useful to be able to look at the stack for the Thread
which is possible right now.
Currently it is not possible to discover where the Fiber was yielded.
Is this something that could be added?
Thread.backtrace example
Should this be in common-ruby?
Could someone move it?
- Project changed from Ruby master to 14
There was some confusion about what I am proposing.
I realised I didn't even explain this!
I would like to be able to see the backtrace of individual Fibers.
This would allow our Celluloid.stack_dump to show each point at which the Fibers are waiting.
- Assignee set to ko1 (Koichi Sasada)
Sorry, I can't understand what is your proposal.
At first example https://gist.github.com/halorgium/f63abf177a96d7113ce3,
there is only "thread.backtrace". However, we have already Thread#backtrace.
p Thread.current.backtrace
#=>
ruby 2.1.0dev (2013-03-18 trunk 39808) [i386-mswin32_100]
["t.rb:2:in backtrace'", "t.rb:2:in
'"]
BTW, please add more specific description.
For example, Rdoc style description for proposed methods
is very welcome because we can understand your proposal
and we (especially I, who is not good English writer)
can copy this document to implemented source code!
When you have many Fibers inside a Thread.
You cannot find where the FIbers are currently suspended.
I propose adding a method to Fiber to return this information.
Calling Fiber.current.backtrace would be equivalent to Thread.current.backtrace due to the nature of Fibers.
Does this help with understanding?
- Description updated (diff)
- Status changed from Open to Assigned
- Subject changed from Support discovering yield state of individual Fibers to google gravity
- Subject changed from google gravity to Support discovering yield state of individual Fibers
- Project changed from 14 to Ruby master
Also available in: Atom
PDF
Like0
Like0Like0Like0Like0Like0Like0Like0Like0Like0Like0Like0