Re: Jenkins build is back to normal : flex-sdk_mustella #177
Hi, I just ran -apollo successfully on the VM. The initial run reports 1055 passing and 68 failing tests. The -failures run reports 0 passing and 0 failures, so something is wrong there... Now moving on to the -mobile tests... EdB On Wed, Jun 19, 2013 at 10:01 PM, Alex Harui aha...@adobe.com wrote: I forgot we added a -mobile option to mini_run.sh. You have to set up a local.properties as specified in the wiki except but leave adl_extras and run_mobile_tests and use_apollo commented out. They will get set by the -mobile option. On 6/19/13 12:49 PM, Erik de Bruin e...@ixsoftware.nl wrote: 20 minutes for AIR. 82 minutes for mobile. Add more for -failures runs if needed. Ok, I'll add these 2 runs to jenkins.sh. Might be a good idea to keep the 'failures.txt' of each completed set (regular + -failures)? I suppose, or you could just bail at whatever run first produces failures that a -failures run won't clean up. The -failures run must also have the -apollo or -mobile option. I got a lot of bitmap failures that I'm looking into now. Which reminded me that if a check-in goes bad and generates a lot of bitmap failures that slows things down (all the bad pngs have to be written to disk). Is there any way to set up jenkins to wait N minutes after a job completes before starting the next job? What would happen if we run every 12 hours but sometimes a job doesn't complete in those 12 hours? I don't think Jenkins will stop one job just to start it again because of a schedule... But I'll read up on it. The current run is 9 hrs, so adding 1.5 hrs still leaves us with plenty of room on 'reasonably clean' runs, right? EdB -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Re: Jenkins build is back to normal : flex-sdk_mustella #177
Hi, I just ran -apollo successfully on the VM. The initial run reports 1055 passing and 68 failing tests. What version of AIR out of interest? Justin
Re: Jenkins build is back to normal : flex-sdk_mustella #177
3.7, I think... how can I be sure? I looked at the 'airsdk.xml' in AIR_HOME and that has airSdk xmlns=http://ns.adobe.com/air/sdk/3.7; EdB On Thu, Jun 20, 2013 at 11:52 AM, Justin Mclean jus...@classsoftware.com wrote: Hi, I just ran -apollo successfully on the VM. The initial run reports 1055 passing and 68 failing tests. What version of AIR out of interest? Justin -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Re: Jenkins build is back to normal : flex-sdk_mustella #177
1055 passing and 68 failing tests. The -failures run reports 0 passing and 0 failures, so something is wrong there... Ah, it looks like what's wrong was me... I didn't put in '-apollo' on the failures run. The full command for failures on the AIR tests is now: sh ./mini_run.sh -apollo -failures Is that correct? EdB -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Re: Jenkins build is back to normal : flex-sdk_mustella #177
Ok, the -mobile target also runs now. I see 3932 passing and 18 failing tests. Now I'll re-run the -apollo target to see if the changes to the -failures command have a positive effect... EdB On Thu, Jun 20, 2013 at 12:26 PM, Erik de Bruin e...@ixsoftware.nl wrote: 1055 passing and 68 failing tests. The -failures run reports 0 passing and 0 failures, so something is wrong there... Ah, it looks like what's wrong was me... I didn't put in '-apollo' on the failures run. The full command for failures on the AIR tests is now: sh ./mini_run.sh -apollo -failures Is that correct? EdB -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Re: Jenkins build is back to normal : flex-sdk_mustella #177
Hi, 3.7, I think... how can I be sure? I looked at the 'airsdk.xml' in AIR_HOME and that has airSdk xmlns=http://ns.adobe.com/air/sdk/3.7; Sound right. Probably need to run the normal SDK tests with FP 11.7 as well. Perhaps it would be good to alternate what version of FP the tests are using ie stock standard 11.1, then 11.7 (latest release) and then 11.8 beta? I noticed Adobe has moved to 11.7 as the extended support release () so perhaps we should target that but make sure it all work on 11.1? http://blogs.adobe.com/flashplayer/2013/05/extended-support-release-updated-to-flash-player-11-7.html Thanks, Justin
Re: Jenkins build is back to normal : flex-sdk_mustella #177
If 11.7 is the new long term support version, why not move to that as the default FP version for Apache Flex? Why stick with 11.1? Also, I think we can squeeze 2 full runs of all three targets in a day. I can set up two sets of tests (e.g. 11.7 and the latests and greatest - currently the 11.8 beta) or run the same set twice (my suggestion would be 11.7). And how about the AIR version? EdB On Thu, Jun 20, 2013 at 1:19 PM, Justin Mclean jus...@classsoftware.com wrote: Hi, 3.7, I think... how can I be sure? I looked at the 'airsdk.xml' in AIR_HOME and that has airSdk xmlns=http://ns.adobe.com/air/sdk/3.7; Sound right. Probably need to run the normal SDK tests with FP 11.7 as well. Perhaps it would be good to alternate what version of FP the tests are using ie stock standard 11.1, then 11.7 (latest release) and then 11.8 beta? I noticed Adobe has moved to 11.7 as the extended support release () so perhaps we should target that but make sure it all work on 11.1? http://blogs.adobe.com/flashplayer/2013/05/extended-support-release-updated-to-flash-player-11-7.html Thanks, Justin -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Re: Jenkins build is back to normal : flex-sdk_mustella #177
Hi, If 11.7 is the new long term support version, why not move to that as the default FP version for Apache Flex? Why stick with 11.1? We don't want to force people to use the very latest, many enterprises have locked down systems and may not be able to move to FP 11.7. And how about the AIR version? Re AIR version it's no so much an issue as above (captive runtime etc). Most people expect the latest to work and we've had a few complaints that the installer get the latest. Thanks, Justin
Re: Jenkins build is back to normal : flex-sdk_mustella #177
Ok, having to pick two, would you rather run against 11.1 and 11.7 (both 'long term support') or against 11.1 and the latest and greatest released version (currently 11.7, but soon 11.8)? For AIR we'll always run against the most recent release (non-beta). EdB On Thu, Jun 20, 2013 at 1:42 PM, Justin Mclean jus...@classsoftware.com wrote: Hi, If 11.7 is the new long term support version, why not move to that as the default FP version for Apache Flex? Why stick with 11.1? We don't want to force people to use the very latest, many enterprises have locked down systems and may not be able to move to FP 11.7. And how about the AIR version? Re AIR version it's no so much an issue as above (captive runtime etc). Most people expect the latest to work and we've had a few complaints that the installer get the latest. Thanks, Justin -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Re: Jenkins build is back to normal : flex-sdk_mustella #177
Hi, Ok, having to pick two, would you rather run against 11.1 and 11.7 (both 'long term support') or against 11.1 and the latest and greatest released version (currently 11.7, but soon 11.8)? +1 for 11.1 and 11.7 from me but other people may think differently. Justin
Re: Jenkins build is back to normal : flex-sdk_mustella #177
Interesting... Ran -apollo, this time got 1059 passing and 68 failing. Note the 4 'extra' passing results, while there are still 68 failing tests... Where did those come from? The -apollo -failures no works, but reports 0 passing and 52 failing... What happened to the remaining 16 failures? I'm going to combine what I have and do a full run for 'main', 'apollo' and 'mobile' combined. That'll take the next 12 hours or so ;-) EdB On Thu, Jun 20, 2013 at 11:23 AM, Erik de Bruin e...@ixsoftware.nl wrote: Hi, I just ran -apollo successfully on the VM. The initial run reports 1055 passing and 68 failing tests. The -failures run reports 0 passing and 0 failures, so something is wrong there... Now moving on to the -mobile tests... EdB On Wed, Jun 19, 2013 at 10:01 PM, Alex Harui aha...@adobe.com wrote: I forgot we added a -mobile option to mini_run.sh. You have to set up a local.properties as specified in the wiki except but leave adl_extras and run_mobile_tests and use_apollo commented out. They will get set by the -mobile option. On 6/19/13 12:49 PM, Erik de Bruin e...@ixsoftware.nl wrote: 20 minutes for AIR. 82 minutes for mobile. Add more for -failures runs if needed. Ok, I'll add these 2 runs to jenkins.sh. Might be a good idea to keep the 'failures.txt' of each completed set (regular + -failures)? I suppose, or you could just bail at whatever run first produces failures that a -failures run won't clean up. The -failures run must also have the -apollo or -mobile option. I got a lot of bitmap failures that I'm looking into now. Which reminded me that if a check-in goes bad and generates a lot of bitmap failures that slows things down (all the bad pngs have to be written to disk). Is there any way to set up jenkins to wait N minutes after a job completes before starting the next job? What would happen if we run every 12 hours but sometimes a job doesn't complete in those 12 hours? I don't think Jenkins will stop one job just to start it again because of a schedule... But I'll read up on it. The current run is 9 hrs, so adding 1.5 hrs still leaves us with plenty of room on 'reasonably clean' runs, right? EdB -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Re: Jenkins build is back to normal : flex-sdk_mustella #177
OK, but a reminder that there will be lots of bitmap compare failures when you switch to 11.7 and resolving them are not my favorite thing to do. On 6/20/13 4:19 AM, Justin Mclean jus...@classsoftware.com wrote: Hi, 3.7, I think... how can I be sure? I looked at the 'airsdk.xml' in AIR_HOME and that has airSdk xmlns=http://ns.adobe.com/air/sdk/3.7; Sound right. Probably need to run the normal SDK tests with FP 11.7 as well. Perhaps it would be good to alternate what version of FP the tests are using ie stock standard 11.1, then 11.7 (latest release) and then 11.8 beta? I noticed Adobe has moved to 11.7 as the extended support release () so perhaps we should target that but make sure it all work on 11.1? http://blogs.adobe.com/flashplayer/2013/05/extended-support-release-update d-to-flash-player-11-7.html Thanks, Justin
Re: Jenkins build is back to normal : flex-sdk_mustella #177
On 6/20/13 5:04 AM, Erik de Bruin e...@ixsoftware.nl wrote: Interesting... Ran -apollo, this time got 1059 passing and 68 failing. Note the 4 'extra' passing results, while there are still 68 failing tests... Where did those come from? Sometimes a SWF just doesn't startup. I have no idea why. But there is no infrastructure to catch that and report how many tests were in it. The -apollo -failures no works, but reports 0 passing and 52 failing... What happened to the remaining 16 failures? This could be an error in the testDir properties of the missing tests. I'm going to combine what I have and do a full run for 'main', 'apollo' and 'mobile' combined. That'll take the next 12 hours or so ;-) EdB On Thu, Jun 20, 2013 at 11:23 AM, Erik de Bruin e...@ixsoftware.nl wrote: Hi, I just ran -apollo successfully on the VM. The initial run reports 1055 passing and 68 failing tests. The -failures run reports 0 passing and 0 failures, so something is wrong there... Now moving on to the -mobile tests... EdB On Wed, Jun 19, 2013 at 10:01 PM, Alex Harui aha...@adobe.com wrote: I forgot we added a -mobile option to mini_run.sh. You have to set up a local.properties as specified in the wiki except but leave adl_extras and run_mobile_tests and use_apollo commented out. They will get set by the -mobile option. On 6/19/13 12:49 PM, Erik de Bruin e...@ixsoftware.nl wrote: 20 minutes for AIR. 82 minutes for mobile. Add more for -failures runs if needed. Ok, I'll add these 2 runs to jenkins.sh. Might be a good idea to keep the 'failures.txt' of each completed set (regular + -failures)? I suppose, or you could just bail at whatever run first produces failures that a -failures run won't clean up. The -failures run must also have the -apollo or -mobile option. I got a lot of bitmap failures that I'm looking into now. Which reminded me that if a check-in goes bad and generates a lot of bitmap failures that slows things down (all the bad pngs have to be written to disk). Is there any way to set up jenkins to wait N minutes after a job completes before starting the next job? What would happen if we run every 12 hours but sometimes a job doesn't complete in those 12 hours? I don't think Jenkins will stop one job just to start it again because of a schedule... But I'll read up on it. The current run is 9 hrs, so adding 1.5 hrs still leaves us with plenty of room on 'reasonably clean' runs, right? EdB -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Re: Jenkins build is back to normal : flex-sdk_mustella #177
On 6/20/13 3:26 AM, Erik de Bruin e...@ixsoftware.nl wrote: 1055 passing and 68 failing tests. The -failures run reports 0 passing and 0 failures, so something is wrong there... Ah, it looks like what's wrong was me... I didn't put in '-apollo' on the failures run. The full command for failures on the AIR tests is now: sh ./mini_run.sh -apollo -failures Is that correct? Yup. Also note that there is a -mobile option that won't require you to muck with the local.properties file. You can setup local.properties once but leave run_mobile_tests and adl_extras commented out and when you run with -mobile it will set those properties for you. EdB -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Re: Jenkins build is back to normal : flex-sdk_mustella #177
FWIW, I think the tests are still running in some earlier namespace. On 6/20/13 2:56 AM, Erik de Bruin e...@ixsoftware.nl wrote: 3.7, I think... how can I be sure? I looked at the 'airsdk.xml' in AIR_HOME and that has airSdk xmlns=http://ns.adobe.com/air/sdk/3.7; EdB On Thu, Jun 20, 2013 at 11:52 AM, Justin Mclean jus...@classsoftware.com wrote: Hi, I just ran -apollo successfully on the VM. The initial run reports 1055 passing and 68 failing tests. What version of AIR out of interest? Justin -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Re: Jenkins build is back to normal : flex-sdk_mustella #177
And if we solve them... will there be as many failures when we go back to 11.1? We were thinking we might run once a day against 11.1 and once a day against 11.7... EdB On Thu, Jun 20, 2013 at 5:12 PM, Alex Harui aha...@adobe.com wrote: OK, but a reminder that there will be lots of bitmap compare failures when you switch to 11.7 and resolving them are not my favorite thing to do. On 6/20/13 4:19 AM, Justin Mclean jus...@classsoftware.com wrote: Hi, 3.7, I think... how can I be sure? I looked at the 'airsdk.xml' in AIR_HOME and that has airSdk xmlns=http://ns.adobe.com/air/sdk/3.7; Sound right. Probably need to run the normal SDK tests with FP 11.7 as well. Perhaps it would be good to alternate what version of FP the tests are using ie stock standard 11.1, then 11.7 (latest release) and then 11.8 beta? I noticed Adobe has moved to 11.7 as the extended support release () so perhaps we should target that but make sure it all work on 11.1? http://blogs.adobe.com/flashplayer/2013/05/extended-support-release-update d-to-flash-player-11-7.html Thanks, Justin -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Jenkins build is back to normal : flex-sdk_mustella #177
See http://localhost:8080/job/flex-sdk_mustella/177/changes
Re: Jenkins build is back to normal : flex-sdk_mustella #177
Cool, we had a clean run!!! Let the games begin ;-) EdB On Wed, Jun 19, 2013 at 11:01 AM, flex.muste...@gmail.com wrote: See http://localhost:8080/job/flex-sdk_mustella/177/changes -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Re: Jenkins build is back to normal : flex-sdk_mustella #177
Hi, Cool, we had a clean run!!! What version of FP was being used? (Windows and 11.1) And you'll probably hate me for bringing this up but did we run the mobile or AIR tests? Thanks, Justin
Re: Jenkins build is back to normal : flex-sdk_mustella #177
11.1, indeed. And I guess not on the AIR and mobile tests... what ant targets/switches make that happen? EdB On Wed, Jun 19, 2013 at 11:41 AM, Justin Mclean jmcl...@apache.org wrote: Hi, Cool, we had a clean run!!! What version of FP was being used? (Windows and 11.1) And you'll probably hate me for bringing this up but did we run the mobile or AIR tests? Thanks, Justin -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Re: Jenkins build is back to normal : flex-sdk_mustella #177
For AIR it should be ./mini_run.sh -apollo tests/apollo For mobile you'll need to swap in a different local.properties before running mini_run on the mobile directory. See the wiki page for how to set that up. -Alex On 6/19/13 2:48 AM, Erik de Bruin e...@ixsoftware.nl wrote: 11.1, indeed. And I guess not on the AIR and mobile tests... what ant targets/switches make that happen? EdB On Wed, Jun 19, 2013 at 11:41 AM, Justin Mclean jmcl...@apache.org wrote: Hi, Cool, we had a clean run!!! What version of FP was being used? (Windows and 11.1) And you'll probably hate me for bringing this up but did we run the mobile or AIR tests? Thanks, Justin -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Re: Jenkins build is back to normal : flex-sdk_mustella #177
Can you give me an indication how long a run of each type takes? EdB On Wednesday, June 19, 2013, Alex Harui wrote: For AIR it should be ./mini_run.sh -apollo tests/apollo For mobile you'll need to swap in a different local.properties before running mini_run on the mobile directory. See the wiki page for how to set that up. -Alex On 6/19/13 2:48 AM, Erik de Bruin e...@ixsoftware.nl javascript:; wrote: 11.1, indeed. And I guess not on the AIR and mobile tests... what ant targets/switches make that happen? EdB On Wed, Jun 19, 2013 at 11:41 AM, Justin Mclean jmcl...@apache.orgjavascript:; wrote: Hi, Cool, we had a clean run!!! What version of FP was being used? (Windows and 11.1) And you'll probably hate me for bringing this up but did we run the mobile or AIR tests? Thanks, Justin -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Re: Jenkins build is back to normal : flex-sdk_mustella #177
This is definitely a great milestone :-) Om On Jun 19, 2013 2:32 AM, Erik de Bruin e...@ixsoftware.nl wrote: Cool, we had a clean run!!! Let the games begin ;-) EdB On Wed, Jun 19, 2013 at 11:01 AM, flex.muste...@gmail.com wrote: See http://localhost:8080/job/flex-sdk_mustella/177/changes -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
Re: Jenkins build is back to normal : flex-sdk_mustella #177
I just tried it on my Mac. 20 minutes for AIR. 82 minutes for mobile. Add more for -failures runs if needed. I got a lot of bitmap failures that I'm looking into now. Which reminded me that if a check-in goes bad and generates a lot of bitmap failures that slows things down (all the bad pngs have to be written to disk). Is there any way to set up jenkins to wait N minutes after a job completes before starting the next job? What would happen if we run every 12 hours but sometimes a job doesn't complete in those 12 hours? -Alex On 6/19/13 7:28 AM, Erik de Bruin e...@ixsoftware.nl wrote: Can you give me an indication how long a run of each type takes? EdB On Wednesday, June 19, 2013, Alex Harui wrote: For AIR it should be ./mini_run.sh -apollo tests/apollo For mobile you'll need to swap in a different local.properties before running mini_run on the mobile directory. See the wiki page for how to set that up. -Alex On 6/19/13 2:48 AM, Erik de Bruin e...@ixsoftware.nl javascript:; wrote: 11.1, indeed. And I guess not on the AIR and mobile tests... what ant targets/switches make that happen? EdB On Wed, Jun 19, 2013 at 11:41 AM, Justin Mclean jmcl...@apache.orgjavascript:; wrote: Hi, Cool, we had a clean run!!! What version of FP was being used? (Windows and 11.1) And you'll probably hate me for bringing this up but did we run the mobile or AIR tests? Thanks, Justin -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl