Tails

#2 Updated by anonym 2015-06-27 08:30:21

• Assignee set to anonym
• Target version set to Tails_1.6

Haven’t verified yet.

#4 Updated by anonym 2015-06-27 09:19:13

• Status changed from New to Confirmed

It worked a few times, but I’ve got this now:

calling as amnesia: xdotool search --name 'Buddy List' windowactivate --sync
call returned: [0, "", ""]
XXX GOT: 
    Then Pidgin successfully connects to the "irc.oftc.net" account
      FindFailed: can not find PidginReconnect.png on the screen.
      Line ?, in File ? (RuntimeError)
      features/pidgin.feature:22:in `Then Pidgin successfully connects to the "irc.oftc.net" account' 

which means that findAny() returned an empty string, which should be impossible (unless the empty string is one of the images it’s looking for, which shouldn’t work either). Or that it returned nil. Note that I didn’t get the same FindFailed error later either. Hmm.

#6 Updated by anonym 2015-06-27 11:40:37

• Assignee changed from anonym to kytv

Ok. I believe we’ve stumbled upon some strange Rjb bug. It seems that Rjb::throw isn’t completely reliable. In waitAny we have return self.findAny(images), and it turns out that in the failing cases the findAny call return nil. It does that because the Rjb::trow in findAny doesn’t kill execution and throw the exception, so its return value, which happens to be nil, is returned instead. Urgh. Worse: sometimes the exception is thrown, but later (from another thread?). This is messy…

This fixes the problem for me. Can you please try, kytv?

--- a/features/support/helpers/sikuli_helper.rb
+++ b/features/support/helpers/sikuli_helper.rb
@@ -36,6 +36,9 @@ for p in package_members
   mod.const_set(class_name, imported_class)
 end

+class FindAnyFailed < RuntimeError
+end
+
 def findfailed_hook(pic)
   STDERR.puts ""
   STDERR.puts "FindFailed for: #{pic}"
@@ -142,8 +145,7 @@ def sikuli_script_proxy.new(*args)
       end
     end
     # If we've reached this point, none of the images could be found.
-    Rjb::throw('org.sikuli.script.FindFailed',
-               "can not find any of the images #{images} on the screen")
+    raise FindAnyFailed.new("can not find any of the images #{images} on the screen")
   end

   def s.waitAny(images, time)
@@ -151,14 +153,13 @@ def sikuli_script_proxy.new(*args)
       loop do
         begin
           return self.findAny(images)
-        rescue FindFailed
+        rescue FindAnyFailed
           # Ignore. We want to retry until we timeout.
         end
       end
     end
   rescue Timeout::Error
-    Rjb::throw('org.sikuli.script.FindFailed',
-               "can not find any of the images #{images} on the screen")
+    raise FindAnyFailed.new("can not find any of the images #{images} on the screen")
   end

   def s.hover_point(x, y)

#7 Updated by kytv 2015-06-27 18:48:20

• Parent task set to Feature #8539

Looking good so far. :) I patched my kytv:test/9517-retry-oftc with Bug #9633#note-3 and it’s working as expected.

We’re inching closer to crossing out Feature #8539 :)

I’m still stress-testing the modified kytv:test/9517-retry-oftc with the above patch, but I’m pretty sure this bug is fixed.

#8 Updated by kytv 2015-06-29 02:11:26

• blocks Feature #9653: Optimize retrying to connect to IRC servers in testsuite using waitAny/existsAny added

#9 Updated by kytv 2015-06-29 02:12:04

• Assignee changed from kytv to anonym

I’m satisfied with your proposed changes.

#10 Updated by kytv 2015-09-04 13:07:28

• QA Check set to Dev Needed

#12 Updated by anonym 2015-09-07 18:18:43

• Status changed from Confirmed to In Progress

Applied in changeset commit:f09d05a75853414ca19f0003050f61f2c613fb1d.

#13 Updated by anonym 2015-09-07 18:20:05

• Assignee changed from anonym to kytv
• % Done changed from 0 to 50
• QA Check changed from Dev Needed to Ready for QA
• Feature Branch set to test/9633-fix-waitAny

Woo, I think I’ve found a proper solution. I only have a theory of why it works (see the commit message) but at this stage that’s good enough for me.

#14 Updated by kytv 2015-09-12 15:46:11

• blocked by deleted (Feature #9653: Optimize retrying to connect to IRC servers in testsuite using waitAny/existsAny)

#15 Updated by kytv 2015-09-12 17:49:11

• blocks Feature #6306: Write tests for I2P added

#16 Updated by kytv 2015-09-12 17:53:05

• Assignee changed from kytv to intrigeri

I’m happy with this fix. Tested and working fine in the branch that I wanted this fix for. :)

#17 Updated by kytv 2015-09-13 16:08:45

• blocks Feature #9653: Optimize retrying to connect to IRC servers in testsuite using waitAny/existsAny added

#18 Updated by kytv 2015-09-14 09:48:24

• Assignee changed from intrigeri to anonym
• QA Check changed from Ready for QA to Dev Needed

Reassigning to anonym based on a discussion on IRC.

With the following code

    on_screen, _ = @screen.waitAny([expected_channel_entry, reconnect_button], 10)
    if on_screen != expected_channel_entry
      raise StandardError.new("Connecting to account #{account} failed.")
    end
    @screen.wait(expected_channel_entry)

wrapped in a function like retry_tor from Bug #10116 (revision 35ee5cf32d009e69b78c31632ad5934668977384) yields the following result when none of the images are found:

When I activate the “I2P” Pidgin account # features/step_definitions/pidgin.rb:293

[log] CLICK on (711,509)

And I close Pidgin’s account manager window # features/step_definitions/pidgin.rb:289

calling as amnesia: xdotool search —name ‘Buddy List’ windowactivate —sync

call returned: [0, “”, “”]

Operation failed (Try 1 of 10) with:

StandardError: Connecting to account I2P failed.

Then Pidgin successfully connects to the “I2P” account # features/step_definitions/pidgin.rb:318

FindFailed: can not find any of the images [“PidginI2PChannelEntry.png”, “PidginReconnect.png”] on the screen

Line ?, in File ? (RuntimeError)

features/i2p.feature:62:in `Then Pidgin successfully connects to the “I2P” account’

And I can join the “#i2p” channel on “I2P” # features/step_definitions/pidgin.rb:365

Scenario failed at time 00:06:59

    
  

#19 Updated by intrigeri 2015-09-15 12:02:42

• Priority changed from Normal to High

(Blocks a high-prio ticket.)

#20 Updated by anonym 2015-09-16 14:44:21

• Assignee changed from anonym to kytv
• QA Check changed from Dev Needed to Ready for QA

I can reproduce it. Indeed, Rjb::throw apparently doesn’t work very well. It seems like Ruby won’t block execution until it has received the Java exception, and in fact it may even get lost completely. I’ve even seen it appear suddenly later, after the scenario has already failed due to another exception being raised. Madness. At least we seem to not have this problem when some Java method we call (via Rjb) throws the exception (e.g. like all those Sikuli methods).

Indeed, it seems we need to introduce a new excetion for the *any() methods. Pushed.

#21 Updated by kytv 2015-09-17 08:59:18

• Assignee changed from kytv to intrigeri

anonym wrote:

> Indeed, it seems we need to introduce a new excetion for the *any() methods. Pushed.

That seemed to solve the problem. Back to intrigeri for his skillful nitpicking.

#22 Updated by intrigeri 2015-09-17 12:28:01

• % Done changed from 50 to 60

I don’t fully understand the code but it doesn’t sound crazy and I’m the third one looking at it. Running what seem to be the two affected features: features/pidgin.feature features/unsafe_browser.feature.

#23 Updated by intrigeri 2015-09-17 13:25:25

• Status changed from In Progress to Fix committed
• % Done changed from 60 to 100
• QA Check changed from Ready for QA to Pass

Merged (commit:b8d2a80)!

#24 Updated by intrigeri 2015-09-17 13:25:59

• Assignee deleted (intrigeri)

#25 Updated by bertagaz 2015-09-22 15:19:33

• blocked by deleted (Feature #9653: Optimize retrying to connect to IRC servers in testsuite using waitAny/existsAny)

#26 Updated by bertagaz 2015-09-22 15:19:44

• Status changed from Fix committed to Resolved
• QA Check deleted (Pass)