#general
- e
enough-analyst-54434
08/04/2015, 2:31 AMYeah - as long as you are sane and have unique struct/service/constant names across thift files sharing a namespace - you;re good - e
enough-analyst-54434
08/04/2015, 2:31 AMI was imagining the old-way of comipling (--recursive from unique roots), might have masked badness in the thrift family of files - e
enough-analyst-54434
08/04/2015, 2:32 AMThe strict, as I found from looking at thrift source-code, only turns on a check to fail structs without field id's declared - e
enough-analyst-54434
08/04/2015, 2:32 AMThat would be fail fast with an obvious message from the thrift comiler - m
mammoth-autumn-2551
08/04/2015, 2:33 AMyep... well don't rush this out on my account, I'll be backporting the chroot fixes onto 0.0.39 for release internally. - e
enough-analyst-54434
08/04/2015, 2:33 AMk - m
mammoth-autumn-2551
08/04/2015, 2:33 AMthis isn't a blocker for us, I was just testing the upgrade while I was immersed in pants-land - m
mammoth-autumn-2551
08/04/2015, 2:34 AMThanks again for you help! - e
enough-analyst-54434
08/04/2015, 2:35 AMYou're welcome, although more and more of the help is fixing issues I personally introduced recently 😕 - a
adorable-hairdresser-18353
08/04/2015, 5:25 PMHey there! I’m running into trouble with the fake options used for testing. It seems that ‘l’ is registered as an option but ‘level’ isn’t. Does anyone know why? - h
happy-kitchen-89482
08/04/2015, 5:25 PMhmmm - h
happy-kitchen-89482
08/04/2015, 5:26 PMthey should both be registered, if not it's a bug. - h
happy-kitchen-89482
08/04/2015, 5:26 PMWhat's the example? - a
adorable-hairdresser-18353
08/04/2015, 5:28 PMwhen I make a call to
in round_engine it breakstask.get_options().leveltest_round_engine - h
happy-kitchen-89482
08/04/2015, 5:28 PMwe have very few options registered with multiple names (that's basically the only one, I think?) so I wouldn't be shocked if there were issues. - h
happy-kitchen-89482
08/04/2015, 5:28 PMbut
works?task.get_options().l - a
adorable-hairdresser-18353
08/04/2015, 5:29 PMwait actually it doesn’t I think I described the problem wrong sorry! - h
happy-kitchen-89482
08/04/2015, 5:30 PMah, in that case maybe the options haven't been registered at all? - h
happy-kitchen-89482
08/04/2015, 5:30 PMjust double-checked the test code and it looks OK to me in terms of treating all aliases equally - a
adorable-hairdresser-18353
08/04/2015, 5:31 PMI was wrong about it being an alias issue. It is really an issue I’m having with the
method in__getattr__fakes.py - a
adorable-hairdresser-18353
08/04/2015, 5:35 PMI’ll get back to you with what I find - h
happy-kitchen-89482
08/04/2015, 5:35 PM
can be tricksy__getattr__ - a
adorable-hairdresser-18353
08/04/2015, 5:36 PMhow so? - e
enough-analyst-54434
08/04/2015, 5:41 PMgenerally recursion issues - e
enough-analyst-54434
08/04/2015, 5:45 PMThe existing
in that test would need expansion for any new global options the round engine accesses in prod codeself.set_options_for_scope('', explain=False) - e
enough-analyst-54434
08/04/2015, 5:46 PMnmm, that's a lie since the test uses self.context() which should be registering all the global options - c
calm-processor-54456
08/04/2015, 6:40 PMhey, i'm with paul at compass, just wanted to make myself known simple smile - b
bored-art-40741
08/04/2015, 6:43 PMWelcome! - e
enough-analyst-54434
08/04/2015, 6:58 PMI've got an un-planned release staged for today here: https://rbcommons.com/s/twitter/r/2569/ - I'm planning to cut it as soon as its dependency is shipped but I can hold off to later in the day if anyone else has fixes they know of in-flight - a
adorable-hairdresser-18353
08/04/2015, 7:32 PM@happy-kitchen-89482: when I call
on a task inside tests I get an instance ofget_options()
which seems to be a wrapper for an empty dictionary? It throws a keyerror for global optionsFakeOptionValues