Automation starting point

(10:15:21 AM) Marco: i'd start with part of system which is easiest to automate. reasons for this will be
(10:16:12 AM) Marco: you want to demonstrate to management that the time you spend on automation is goign to save time once u have it set up and running
(10:16:22 AM) Marco: to best prove automation has value
(10:16:36 AM) Marco: automate something easy. so that it will run reliably.
(10:17:06 AM) Marco: if you can get that to work..and convince some manager or developers you will have more support from them when or if the time comes to automate some harder parts of system
(10:17:07 AM) Grace: if they already decide to start automation anyway, is still the same ansers
(10:17:57 AM) Marco: yes i would still say same answer. the reason for this is i have gone into a several jobs where management was totally in support of automation, they had no clue, all they thought was we need to automate it will save us
(10:18:44 AM) Marco: n what happened was. when i explained that to setup automation it would take some time, or that developers might need to make changes to the src code to make it easier for automated test tools to hook into it..
(10:18:48 AM) Marco: thats when it started to fall apart
(10:18:56 AM) Grace: ah...
(10:19:24 AM) Marco: you want to avoid as much as possible the mentality by managers that 'automation will cure all our testing problems' without having to invest any time or effort in actually doing automation
(10:19:56 AM) Marco: so for these reasons based on my experience. start with somehting easy or small..
(10:20:06 AM) Marco: dont try to automate any parts of system that are unstable
(10:20:26 AM) Marco: if you can demonstrate that the tests run reliably 9 times out of 10
(10:20:28 AM) Marco: good
(10:21:16 AM) Marco: if u have tests that are running and failing for a bunch of reasons like network timeouts, or stuff that is outside your control then ppl will see that result. they will think. automation = waste of time
(10:22:19 AM) Marco: also my final comment. you won't convince everyone. but if you can convince even 1 or 2 of senior devs n someone in management and show them a good result from a small set of tests, you will have more support to automate the rest

0 comments:

Post a Comment