Datasets:

showvikdbz

/

code-reviews

like 0

Because this change was done few commits ago. We could ignore and have one but if someone has saves of this 0.9.5 release version they're screwed.

now you must update heads instead of head because of

Is this name clear? Maybe `BufferedDAGAdder` (yuck!)?

need to initialize put offset here or maybe as an initializer in flushparams

Is requiring v3 an option? A package manager for HPC that forbids use of vector registers is kinda ridiculous... In any case, the point of this PR is mostly not to build for x86-64-v4 by accident.

We do what now? Oof .. nothing to fix in this PR, but sending fields like this optional is .. euh .. weird.

please remove the white spaces

prob does not need to be imported as btypes

i am good with root secret

what the fuck is a job controller

hell even the tooltips change so people can figure out which way works best for them

If you want to get help, naming your classes and variables right is the first step : it makes your code understandable to others, and even to you, because frankly, I doubt even you can make sense out of this. > by methodology in my country entities are plural and singular variables. You should fix your naming and more importantly, quit this kind of dubious nonsense if you want some help.

really make a no op a static method this is just silly

yeah but my compose function is going to look pretty stupid as a part of this patch if i do not use it your suggestion is definitely a much better cleanup though

does "It's a little cleaning robot. This one appears to have it's cleaning solutions replaced with a weird goo. It also appears to have it's targeting protocols overridden..." sound good?

how about just return job jobstate in self completed job states

You broke the fuck out of the dme

Damn. Thought i got all those. EDIT: Fixed now.

Apparently it just "works", I assume it uses potency or some dumb shit, food code scares me

globals yml

I think I need an abstract PDO DSN builder. It's getting a bit ridiculous hard-coding all these various configurations into one class.

Wtf github, it didn't show any new comment when I was deleting :(

``` /local/EB/build/eb-5ELu_U/bazel_x5LRfhQ7/out/execroot/io_bazel/bazel-out/host/bin/external/com_google_protobuf/protoc: /lib64/libstdc++.so.6: version `CXXABI_1.3.8' not found (required by /local/EB/build/eb-5ELu_U/bazel_x5LRfhQ7/out/execroot/io_bazel/bazel-out/host/bin/external/com_google_protobuf/protoc ``` Well crap, usual nonsense issues here that i was hoping to skip for once.

any reason to have this stored in the and renderers instead of the base class

destroy is doing this when removing all children see innernode removechildren this line becomes a no op anyway since we removed all children out of the clusterlist

Suckers are easy to miss, aren't they? :)

can we just use pid namespaces to not have to rely on process group ids if you kill the init process inside a pid namespace everything else dies

I'm 99% sure that it would, but you would need to do src = null before the spawn. (Also it would still fuck up GC on the target I believe but whatever that rarely would happen)

Hot damn. I thought for sure soft shadows could never be done. It's official, you guys can do anything now. All hail Lord FrozenFish!

lines 421 and 422 look really fucked up too

Chaplain McGuffin hits you with the FUCKNUKEOPS.

only parse drv info mock assert called once with must be located in the block with because it mocked via context manager the remaining mocks at level test case can be outside the block with

It's a goddamn comment

What the hell would I do then?

looks like you are missing a newline here

i would prefer not to have the except exception block on this test or any of them if an exception is going to make a test fail just let it go through and fail instead of throwing away the type message line number and stack trace basically next time i am editing the ring builder and i make a stupid typo and get a nameerror i want to see nameerror in my test output so i can go fix my mistake

oh shit wait it is, I thought that was stupid

you do not need if here because default value is os service default

only parse drv info mock assert called once with must be located in the block with because it mocked via context manager the remaining mocks at level test case can be outside the block with

3 space indent is evil since it looks almost right in both directions :)

this is probably a dumb and pointless opinion but admin py does not give me the idea of a cli module as this is i personally prefer keystone approach where the admin mgmt module is just called cli py

you are mixing import styles and generally it is preferred to import the whole module and reference parts of that module with the name of the module included i e reporting reporter rather than reporter it tells the reader immediately where that code came from remove both reporter and testfailure imports and change your references below to use reporting reporter and reporting testfailure

Sure it could be templated, but least of my concerns. The 26 parameters cannot be duplicated all over the place. This is an absurd function signature. Why are we implementing C-APIs, when we don't want to support C?

let s all be clear here all of this is a hack we re in the grimy underside of python here

region code may contain digits and symbol like latam es what about just

minor for start end in self netw

@cielf "what if your nonsense bank names are all before "DEFAULT" " I'd have to look into this tomorrow. But I do want to ask, is it worth it to test pagination here? It seems like this is turning into a bit of a tornado as far as the tests are concerned.

ds browser should be property of the datastore

because i am dumb will do

Weird, no idea where that came from, it'll be fixed in a second :)

5u omnizine after each hit is ridiculous.

dbg should be false to avoid log spew

thinking about killing pkgtype altogether see the non inline comments on the review and share

bad indentations

What? !recoil will be 0 if there is recoil and 1 if there is 0 recoil, it will never be higher than eighty, just use <, what the hell?

execution fail here you will need to enclose the string with double quotes also please fix this one as well your you are

die mixin die

sorry i was stupid i thought this disables memcache caching so please ignore me

why do not you need a slow path in case of null for this one

damn, hopefully my programming skills are better than my knowledge of the alphabet

this is useless if you use a boolean

@SharkyKZ I get an error `1054 Unknown column 'featured_up, featured_down' in 'field list'` when testing your suggested change and opening the article for edit. The swapping of the dates is rubbish anyway, I'll leave that away. Other sanity checks seem already to be done.

this comment might be not be needed as still makes no sense here anymore nit e

stupid question should we be explicit that we are not preserving order here

yeah this comment is worse than unhelpful comment removed

Useful (this sentence is weird)

this is useless as these types already have keystone service autorequire for example keystone endpoint autorequire service do keystone end

huge nit you do not need a colon after on

aha overlooked that why not add for the stupid like me the decrypter is of course needed to handle all get requests

Dang, what does this do again?

nit data get with subnet would work just fine since none is a false value

it s a place to put any code that will be executed

allocating a new calendar object every time seems wasteful

delete openstack baremetal

i think we can kill indentation level by making this check part of the if structure

i have just learned that i told a stupid thing yes i will change this to wait thanks for this catch

need to preserve padding and style border right px solid white here

Disgusting, delete moods now

subnet self make subnet fmt or self fmt

I wonder how often this comes up, I don't want to bloat this if it's rare enough to not be an issue. s a valid point tho shit sucks

wtf is this gay shit lmao

so if i do enable plane and disable plane shortly after that before a vblank interrupt lands am i going to mess up my drm vblank refcount

\\* and / are commutative, and this isn't a case where floating point shit could butt bite us. So even if it wasn't L->R we'd be ok. I can do it if you'd like but I think it's fine.

i think you need to indent the s to make them recognized as bullets

nit deleted binding of the snat

I am a moron but these feel wrong

fk to which table i have looked and do not see a table for this

i will kill this ternary operator and do this correctly with an if

I really hate this naming scheme you've got going on here

i am with jay on this one i prefer checking conf options on driver init and barfing if something s wrong there are dumb admins but also ones that just make a typo here and there

gfx kplaceholdercolor instead of black then you probably do not need the comment

i would change it to this all puppet openstack classes and resources should be declared in granular tasks directly not via openstack module

that would apply my cflags arm to all other archs use local asflags arm instead

Yeah it is a bit clumsy, but for now it works the gurus like yourself and oldmcg will be able to sort the code properly, just trying to give back to a community thats helped me :)

Choppy as fuck, but best I could get to actually work.

who the fuck ruins their lasagna with aubergines?

edp debugfs dir points at garbage now

yo momma so fat, her ghost drifts through space

Because these are only main jobs that are supposed to sit and fuck beside the cloner.

afaiu this warning is not relevant anymore