#general
- e
enough-analyst-54434
09/22/2015, 10:17 PMWe've previously gone down this road and backed out - e
enough-analyst-54434
09/22/2015, 10:17 PMSo motivation for why this is needed at the target level need to be provided in the RB description - e
enough-analyst-54434
09/22/2015, 10:18 PMWe already had python test timeouts in the past - e
enough-analyst-54434
09/22/2015, 10:19 PMIn short - my guess is you are on the test team! - e
enough-analyst-54434
09/22/2015, 10:19 PMCommunication flow has always been suboptimal - l
limited-terabyte-34878
09/22/2015, 10:19 PMYes! I am on CI team. good guess. - l
limited-terabyte-34878
09/22/2015, 10:19 PM😃 - e
enough-analyst-54434
09/22/2015, 10:19 PMNot hard - w
witty-crayon-22786
09/22/2015, 10:20 PMitay has suggested that allowing targets to configure their own timeouts would be good. - e
enough-analyst-54434
09/22/2015, 10:20 PMJust asking for some public talk on why - w
witty-crayon-22786
09/22/2015, 10:20 PMi shrug, but - e
enough-analyst-54434
09/22/2015, 10:20 PMPlease just motivate all this to the external world with examples - l
limited-terabyte-34878
09/22/2015, 10:20 PMthe description was terse because i thought i was submitting for only for an internal review with people who’ve already seen the reasoning etc. sorry about that, I’ll provide a more in depth description. - w
witty-crayon-22786
09/22/2015, 10:20 PM@limited-terabyte-34878: would be worth doing a full motivation/explanation on there probably. or open up your doc - e
enough-analyst-54434
09/22/2015, 10:21 PMYeah - I'd watch out on internal RB 1st - e
enough-analyst-54434
09/22/2015, 10:21 PMIf you had kept this internal, you might have done $$$ of work - e
enough-analyst-54434
09/22/2015, 10:21 PMto get thrown out on open-up - e
enough-analyst-54434
09/22/2015, 10:21 PMIts a danger - l
limited-terabyte-34878
09/22/2015, 10:21 PMyeah that’s true john - e
enough-analyst-54434
09/22/2015, 10:24 PMMy guess is the motivation is again - seperate teams, and easier modification of bad actors without actually mucking in code - which is riskier, ie test team adds a timeout parameter to a BUILD file - maybe via automation, maybe not - either way less risky and quicker than having a dev modify their test code to add a timeout to a particular test method function - e
enough-analyst-54434
09/22/2015, 10:24 PMThat separation isn't necessarily bad - e
enough-analyst-54434
09/22/2015, 10:24 PMbut its a thing and should be called as a shot, etc etc - l
limited-terabyte-34878
09/22/2015, 10:26 PMOur issue is that our CI jobs for core libraries are exceeding the global timeout that we have set for CI jobs because they have to run a lot of builds and tests. Rather than increasing the timeout we figure it makes sense to push the timeout down to the lowest level to catch things that hang more quickly/reactively. I’ll put more details into the RB - e
enough-analyst-54434
09/22/2015, 10:27 PMBut when you do so - be clear this is not in fact the lowest level - e
enough-analyst-54434
09/22/2015, 10:27 PMThe lowest level is @Test(timeout=) etc - e
enough-analyst-54434
09/22/2015, 10:27 PMYou want up a level - in BUILD, presumably for the reasons above - e
enough-analyst-54434
09/22/2015, 10:27 PMIE these failures go on your teams budget - e
enough-analyst-54434
09/22/2015, 10:27 PMAnd you dont want to mess with another teams code - e
enough-analyst-54434
09/22/2015, 10:27 PMBecause of risk or expedience or what have you - e
enough-analyst-54434
09/22/2015, 10:29 PMI know twitter politics to some degree and assume they have not shifted dramatically