| *** exploreshaifali has quit IRC | 00:18 | |
| *** echevemaster has joined #openstack-zaqar | 00:38 | |
| *** amitgandhinz has quit IRC | 00:42 | |
| openstackgerrit | Jeffrey Zhang proposed a change to openstack/zaqar: Add enum34 requirement https://review.openstack.org/127828 | 01:28 |
|---|---|---|
| *** cpallares has quit IRC | 01:31 | |
| *** cpallares has joined #openstack-zaqar | 01:34 | |
| *** jeffrey4l has quit IRC | 01:55 | |
| *** jeffrey4l has joined #openstack-zaqar | 02:08 | |
| *** X019 has quit IRC | 02:11 | |
| openstackgerrit | Jeffrey Zhang proposed a change to openstack/zaqar: Sync the zaqar.conf.sample file https://review.openstack.org/128158 | 02:43 |
| *** cpallares has quit IRC | 02:54 | |
| *** flwang has quit IRC | 02:54 | |
| *** flwang has joined #openstack-zaqar | 03:07 | |
| *** cpallares has joined #openstack-zaqar | 03:09 | |
| openstackgerrit | A change was merged to openstack/zaqar: Fix no links property in flavor list response https://review.openstack.org/123626 | 03:22 |
| *** jeffrey4l has quit IRC | 03:44 | |
| *** jeffrey4l has joined #openstack-zaqar | 04:05 | |
| *** bradjones has quit IRC | 05:40 | |
| *** vipul has quit IRC | 05:40 | |
| *** bradjones has joined #openstack-zaqar | 05:46 | |
| *** vipul has joined #openstack-zaqar | 05:46 | |
| *** echevemaster has quit IRC | 05:57 | |
| *** jeffrey4l has quit IRC | 06:31 | |
| *** sgotliv has quit IRC | 06:35 | |
| *** jeffrey4l has joined #openstack-zaqar | 06:43 | |
| *** exploreshaifali has joined #openstack-zaqar | 06:48 | |
| *** exploreshaifali has quit IRC | 06:52 | |
| *** jeffrey4l has quit IRC | 06:59 | |
| *** flaper87|afk is now known as flaper87 | 07:03 | |
| *** jeffrey4l has joined #openstack-zaqar | 07:08 | |
| *** sgotliv has joined #openstack-zaqar | 07:36 | |
| openstackgerrit | Jeffrey Zhang proposed a change to openstack/zaqar: Fix the exception raised when using the nosetests run test https://review.openstack.org/128196 | 07:38 |
| openstackgerrit | Jeffrey Zhang proposed a change to openstack/zaqar: Fix the exception raised when running test by nose https://review.openstack.org/128196 | 07:41 |
| *** exploreshaifali has joined #openstack-zaqar | 07:53 | |
| exploreshaifali | flaper87: all the configs ofmongodb comes from https://github.com/openstack/zaqar/blob/master/zaqar/queues/storage/mongodb/options.py | 08:01 |
| exploreshaifali | flaper87: since we need to seperate the configs, I don't understand what exactly should be done. | 08:16 |
| exploreshaifali | options.py as per I understood is defining configs for mongodb | 08:17 |
| exploreshaifali | so that file should be intact | 08:17 |
| exploreshaifali | it is? | 08:17 |
| *** sgotliv has quit IRC | 08:50 | |
| *** sgotliv has joined #openstack-zaqar | 09:02 | |
| *** jeffrey4l has quit IRC | 09:20 | |
| *** jeffrey4l has joined #openstack-zaqar | 09:31 | |
| *** njirap has joined #openstack-zaqar | 10:26 | |
| *** sgotliv has quit IRC | 10:32 | |
| *** bradjones has quit IRC | 10:33 | |
| *** bradjones has joined #openstack-zaqar | 10:33 | |
| *** sgotliv has joined #openstack-zaqar | 10:41 | |
| openstackgerrit | A change was merged to openstack/python-zaqarclient: Updated from global requirements https://review.openstack.org/127814 | 10:56 |
| *** malini has joined #openstack-zaqar | 11:51 | |
| openstackgerrit | Flavio Percoco proposed a change to openstack/python-zaqarclient: Setup developer docs for zaqarclient https://review.openstack.org/127171 | 11:57 |
| openstackgerrit | Flavio Percoco proposed a change to openstack/python-zaqarclient: Extend some docstrings with useful information https://review.openstack.org/127227 | 11:57 |
| openstackgerrit | Flavio Percoco proposed a change to openstack/python-zaqarclient: Add docs for `Client` instances https://review.openstack.org/127182 | 11:57 |
| openstackgerrit | Flavio Percoco proposed a change to openstack/python-zaqarclient: Add reference docs for latest recommended client https://review.openstack.org/127215 | 11:57 |
| *** malini has quit IRC | 11:57 | |
| *** julienvey has quit IRC | 12:10 | |
| *** jchai has joined #openstack-zaqar | 12:57 | |
| *** njirap has quit IRC | 13:07 | |
| flaper87 | jeffrey4l: zhiyan kgriffs|afk malini|afk sebasmagri cpallares exploreshaifali flwang guys and girls, just a heads up. Please, write your RFCs for Kilo in the next couple of weeks (before the summit) and help reviewing the existing ones. | 13:08 |
| flaper87 | We've gotta get our plan for kilo sorted out | 13:08 |
| exploreshaifali | flaper87: I don't know what a kilo mean? | 13:17 |
| flaper87 | exploreshaifali: hehe, that's the name of the next OpenStack release | 13:17 |
| flaper87 | the one you'll be working on | 13:17 |
| *** mpanetta has joined #openstack-zaqar | 13:17 | |
| exploreshaifali | it is Juno? | 13:17 |
| flaper87 | nope, Juno is going to be release this week | 13:18 |
| exploreshaifali | flaper87: yes 16-oct | 13:18 |
| flaper87 | exploreshaifali: exactly, the next one is called Kilo | 13:18 |
| flaper87 | and that's the one you'll be hacking on | 13:18 |
| exploreshaifali | ok :) | 13:18 |
| zhiyan | Sounds cool, will do | 13:19 |
| zhiyan | thanks for reminder ^ | 13:19 |
| *** njirap has joined #openstack-zaqar | 13:20 | |
| flaper87 | These are the RFCs we have so far: https://review.openstack.org/#/q/status:open+project:openstack/zaqar-specs,n,z | 13:20 |
| *** yfujioka has quit IRC | 13:22 | |
| exploreshaifali | zhiyan: please can you help me a bit what we are suppose to do? | 13:22 |
| flaper87 | exploreshaifali: https://github.com/openstack/zaqar-specs | 13:23 |
| flaper87 | you need to propose a patch similar to https://review.openstack.org/#/c/126531/ | 13:24 |
| flaper87 | That is, you need to copy the template.rs into specs/kilo/ | 13:24 |
| flaper87 | with the name of your spec file and edit it | 13:24 |
| flaper87 | exploreshaifali: your spec should contain the info of what you're planning to do with the data planes | 13:24 |
| flaper87 | data/control | 13:24 |
| flaper87 | why that change is needed etc | 13:24 |
| flaper87 | The template has all the questions that need to be answered | 13:25 |
| flaper87 | :) | 13:25 |
| flaper87 | it should be pretty straightforward | 13:25 |
| exploreshaifali | flaper87: ok, Thanks!! | 13:25 |
| * flaper87 goes and writes yet-another-spec | 13:25 | |
| openstackgerrit | A change was merged to openstack/zaqar-specs: Template for new storage driver's addition https://review.openstack.org/125658 | 13:26 |
| *** sriram has joined #openstack-zaqar | 13:27 | |
| *** exploreshaifali has quit IRC | 13:30 | |
| openstackgerrit | Flavio Percoco proposed a change to openstack/zaqar-specs: Propose a drawbacks section https://review.openstack.org/128277 | 13:33 |
| *** flaper87 has quit IRC | 13:35 | |
| *** malini has joined #openstack-zaqar | 13:48 | |
| *** flaper87|afk has joined #openstack-zaqar | 14:09 | |
| openstackgerrit | Flavio Percoco proposed a change to openstack/zaqar-specs: Make FIFO guarantee optional https://review.openstack.org/125986 | 14:12 |
| *** flaper87|afk is now known as flaper87 | 14:12 | |
| *** flaper87 has quit IRC | 14:14 | |
| *** flaper87 has joined #openstack-zaqar | 14:14 | |
| *** amitgandhinz has joined #openstack-zaqar | 14:15 | |
| *** anna_ has joined #openstack-zaqar | 15:04 | |
| *** jchai is now known as jchai_afk | 15:26 | |
| *** jchai_afk is now known as jchai | 15:27 | |
| *** sgotliv has quit IRC | 15:33 | |
| *** riveter has joined #openstack-zaqar | 16:09 | |
| *** kgriffs|afk is now known as kgriffs | 16:34 | |
| *** flaper87 is now known as flaper87|afk | 16:38 | |
| *** sriram has quit IRC | 16:46 | |
| *** sriram1 has joined #openstack-zaqar | 16:47 | |
| *** njirap has quit IRC | 17:07 | |
| *** exploreshaifali has joined #openstack-zaqar | 17:08 | |
| *** anna_ has quit IRC | 17:13 | |
| *** anna_ has joined #openstack-zaqar | 17:25 | |
| *** jchai is now known as jchai_afk | 17:41 | |
| *** sgotliv has joined #openstack-zaqar | 18:03 | |
| *** sebasmagri has quit IRC | 18:06 | |
| *** sebasmagri has joined #openstack-zaqar | 18:08 | |
| *** jchai_afk is now known as jchai | 18:25 | |
| *** earnThis has joined #openstack-zaqar | 19:28 | |
| exploreshaifali | kgriffs, I am trying to understand https://etherpad.openstack.org/p/zaqar-scratch please can you help me in it | 19:33 |
| kgriffs | sure | 19:34 |
| kgriffs | let me add an example showing what was decided | 19:34 |
| *** bradjones has quit IRC | 19:36 | |
| cpallares | flaper87|afk: What are RFCs? | 19:38 |
| kgriffs | exploreshaifali: does that help? | 19:38 |
| exploreshaifali | kgriffs, very fist our objective is to separate data and control plane of storage layer | 19:38 |
| exploreshaifali | and for that first task is to split the configs | 19:39 |
| kgriffs | kk | 19:39 |
| exploreshaifali | kgriffs, these are the configs of datastore right? | 19:39 |
| kgriffs | so, that example I put right under "config options notes" is the schema flaper87 and I came up with | 19:40 |
| kgriffs | whereas before you had just "storage" | 19:40 |
| kgriffs | now that is split into | 19:41 |
| kgriffs | "message_store" | 19:41 |
| kgriffs | and | 19:41 |
| kgriffs | "management_store" | 19:41 |
| exploreshaifali | my question is how can we divide the configs of a datastore? | 19:41 |
| exploreshaifali | datastore configs means the configurations of datastore with which our applications will going to interact | 19:42 |
| exploreshaifali | and these config options will remain same always, their values may change | 19:43 |
| exploreshaifali | kgriffs, am I able to put my question clearly? | 19:43 |
| kgriffs | ok, let me see if I understand the question | 19:44 |
| kgriffs | today we have a single module that defines two classes | 19:44 |
| exploreshaifali | yes | 19:45 |
| kgriffs | DataDriver | 19:45 |
| kgriffs | dn | 19:45 |
| kgriffs | and | 19:45 |
| kgriffs | ControlDriver | 19:45 |
| exploreshaifali | yes | 19:45 |
| kgriffs | but, we use the same config section from the INI file to configure both | 19:45 |
| kgriffs | we just pick which one to load, depending on where pooling is enabled | 19:46 |
| kgriffs | (see also bootstrap.py) | 19:46 |
| kgriffs | actually | 19:46 |
| kgriffs | bootstrap loads a meta-driver from the pooling module | 19:46 |
| kgriffs | just a sec | 19:47 |
| exploreshaifali | https://github.com/openstack/zaqar/blob/master/zaqar/queues/bootstrap.py | 19:48 |
| kgriffs | ah, right | 19:48 |
| kgriffs | now I remember | 19:48 |
| kgriffs | ok, so it loads that meta-driver or "proxy" driver, which proxies requests to one of several true data drivers - basically acting as a router | 19:49 |
| kgriffs | but to construct that, it calls the self.control property | 19:49 |
| kgriffs | and that is what loads up the control driver by setting control_mode to true | 19:49 |
| kgriffs | ok, so instead of using the same config section to load both "storage" and "control" drivers, it should use unique sections, as shown on the etherpad | 19:50 |
| exploreshaifali | in the etherpad we are spliting the configs, right | 19:50 |
| kgriffs | furthermore, in bootstrap, it seems like the "storage" property should be renamed to "message_store" and likewise for "control" ==> "management_store" to be consistent with the new config | 19:51 |
| kgriffs | exploreshaifali: does that make sense? | 19:52 |
| exploreshaifali | we are spliting configs of what? | 19:53 |
| kgriffs | so, before we had a single config section | 19:53 |
| kgriffs | [drivers:storage:name-of-driver] | 19:53 |
| kgriffs | now we have two sections | 19:53 |
| kgriffs | [drivers:message_store:name-of-message-store-driver] | 19:54 |
| exploreshaifali | and each section will describe configs of their own datastores | 19:54 |
| kgriffs | [drivers:management_store:name-of-management-store-driver] | 19:54 |
| kgriffs | exploreshaifali: yeah | 19:54 |
| exploreshaifali | kgriffs, basically these configs describe what? | 19:55 |
| exploreshaifali | details to connect with datastore? | 19:55 |
| kgriffs | we could, for example, decide to keep the sqlalchemy management_store driver, but remove the message_store one. But I expect in most cases for each supported backend we will include 2 drivers. | 19:56 |
| *** anna_ is now known as XX019 | 19:56 | |
| kgriffs | exploreshaifali: ah, yes | 19:56 |
| *** XX019 is now known as X019 | 19:56 | |
| kgriffs | have a look at the example config | 19:56 |
| kgriffs | all of them have a connection string as a URI | 19:56 |
| kgriffs | and then they may have additional driver-specific options | 19:56 |
| exploreshaifali | so with this we will modify options.py | 19:57 |
| kgriffs | for example, SSL settings, which database to select, socket timeouts and sleeps, etc. | 19:57 |
| kgriffs | exploreshaifali: yes | 19:58 |
| kgriffs | but... | 19:58 |
| kgriffs | I think it may be a good idea to actually put the drivers in separate dirs | 19:58 |
| kgriffs | we have queues/storage | 19:59 |
| exploreshaifali | correct | 19:59 |
| kgriffs | I was thinking, perhaps we should add | 19:59 |
| kgriffs | queues/management_store | 19:59 |
| kgriffs | and rename the other to | 19:59 |
| exploreshaifali | then in store we should have data dir and one control dir | 19:59 |
| kgriffs | queues/message_store | 19:59 |
| exploreshaifali | yes that can be done or I was thinking something like | 20:00 |
| kgriffs | exploreshaifali: oh, that's another way to do it | 20:00 |
| kgriffs | queues/storage/message | 20:00 |
| kgriffs | queues/storage/managment | 20:00 |
| exploreshaifali | exactly :) | 20:00 |
| exploreshaifali | so I guess the second on would be more better as it is under storage dir? | 20:01 |
| kgriffs | so then management would have it's own driver and options | 20:01 |
| exploreshaifali | and message will have its own driver and options | 20:01 |
| kgriffs | right | 20:02 |
| *** malini has left #openstack-zaqar | 20:02 | |
| kgriffs | so, first step is to make a patch | 20:02 |
| riveter | wouldn't that involve a lot of duplicated code? | 20:02 |
| exploreshaifali | riveter, I was also thinking the same | 20:03 |
| kgriffs | what would we duplicate? | 20:03 |
| kgriffs | oh, I see | 20:03 |
| exploreshaifali | but before that I more thing need to be clear | 20:03 |
| kgriffs | there will be overlap in many of the options | 20:03 |
| exploreshaifali | what about rest of the files in storage dir? | 20:04 |
| exploreshaifali | like base.py, errors.py | 20:04 |
| kgriffs | although, there will be some options that only apply to message vs management | 20:04 |
| exploreshaifali | they will remain in storage dir only? | 20:04 |
| riveter | so would it make more sense to keep the current files & just make subclasses for the drivers? | 20:04 |
| kgriffs | mmm, not exactly | 20:05 |
| kgriffs | take the case of base.py | 20:05 |
| kgriffs | it defines DriverBase | 20:06 |
| riveter | ya | 20:06 |
| kgriffs | which should be inherited | 20:06 |
| kgriffs | but | 20:06 |
| *** jchai is now known as jchai_afk | 20:06 | |
| kgriffs | it has other stuff in there that is specific to either message or management | 20:06 |
| kgriffs | those things would be moved down on level in the directory tree | 20:06 |
| kgriffs | and then used in their respective contexts | 20:07 |
| kgriffs | wrt options | 20:07 |
| kgriffs | hmm | 20:07 |
| exploreshaifali | yes | 20:07 |
| * kgriffs thinking | 20:07 | |
| riveter | but there is a fair amount of common code too, no? | 20:07 |
| kgriffs | yeah | 20:07 |
| exploreshaifali | common code will remain in storage dir itself | 20:08 |
| exploreshaifali | and specific code will move to their specific dirs | 20:08 |
| kgriffs | what we will have to do is identify the common bits, and be able to reuse them at lower (more specific) levels/modules in the dir tree | 20:08 |
| exploreshaifali | exactly | 20:08 |
| riveter | this is a substantial patch we're talking about | 20:09 |
| kgriffs | yep | 20:09 |
| kgriffs | the problem is the two drivers were not properly decoupled to begin with and now we have technical debt to clean up | 20:09 |
| kgriffs | btw, for the sake of reusing code/options | 20:10 |
| riveter | can it be broken down into smaller patches? | 20:10 |
| kgriffs | I think the dir will have to look more like | 20:10 |
| kgriffs | queues/storage/mongodb/management | 20:10 |
| kgriffs | queues/storage/mongodb/messages | 20:10 |
| kgriffs | so that management and messages can share all things mongo, which would naturally live directly under ueues/storage/mongodb | 20:11 |
| kgriffs | riveter: yes, I think it should be | 20:11 |
| exploreshaifali | kgriffs, yes that could be done | 20:12 |
| kgriffs | the first patch would be to reorganize the directory structure but not split things up yet. | 20:12 |
| kgriffs | (as far as possible) | 20:12 |
| kgriffs | so, for starters, there would just be a queues/storage/mongodb/messages | 20:12 |
| kgriffs | it would still contain both classes in the same driver file, share options, etc. | 20:13 |
| kgriffs | idk, that may or may not provide a natural breakdown of the work. | 20:13 |
| kgriffs | but the idea is, let's try to break it down into steps | 20:13 |
| riveter | who's working on this? | 20:14 |
| exploreshaifali | riveter, I am looking forward to work on this | 20:14 |
| exploreshaifali | :) | 20:14 |
| kgriffs | exploreshaifali: btw, I thought of something else we will have to figure out | 20:15 |
| kgriffs | backwards-compat | 20:15 |
| exploreshaifali | okay | 20:15 |
| kgriffs | i mean, really just from a config standpoint | 20:15 |
| kgriffs | it may be ok just to tell people they need to update their configs before deploying this new version | 20:15 |
| kgriffs | exploreshaifali: you might want to get flaper87's thoughts on it | 20:16 |
| exploreshaifali | yeah | 20:16 |
| exploreshaifali | kgriffs, so if we talk about config spliting only | 20:16 |
| exploreshaifali | for which I started discussion | 20:16 |
| kgriffs | personally, I think it would be OK to ask people to update their configs since it should be pretty easy to do. we could even give them a migration script that helps. but we should ask around | 20:17 |
| kgriffs | exploreshaifali: ok, back to config splitting | 20:17 |
| exploreshaifali | how can I identify which part is for management and which is for storage | 20:17 |
| kgriffs | most config options will be identical (just copy them) | 20:18 |
| kgriffs | let me check | 20:18 |
| exploreshaifali | like in python files, I can clearly identify this is for driver and this code is for data plane | 20:18 |
| exploreshaifali | but here getting very less | 20:18 |
| kgriffs | ok, so for mongo I think both use the same options except for "partitions" | 20:18 |
| kgriffs | "partitions" is only used by the data (AKA message_store) driver | 20:19 |
| kgriffs | is that the nature of your question? | 20:20 |
| exploreshaifali | yes, a bit of | 20:20 |
| exploreshaifali | so you mean that full options.py will be same | 20:20 |
| kgriffs | except for that one field | 20:20 |
| exploreshaifali | except MONGODB_GROUP = 'drivers:storage:mongodb | 20:20 |
| kgriffs | yeah, that too | 20:20 |
| kgriffs | so under queues/storage/mongodb/options.py | 20:21 |
| kgriffs | you would have some variable that declares the common options, which will be everything except 'partitions' | 20:21 |
| exploreshaifali | kgriffs, which field will vary other than MONGODB_GROUP = 'drivers:storage:mongodb' | 20:21 |
| kgriffs | well, what I would do is in that "base" options.py | 20:22 |
| exploreshaifali | kgriffs, what exactly you mean by partitions? | 20:22 |
| kgriffs | don't declare a group of a _config_options | 20:22 |
| kgriffs | exploreshaifali: that is the name of one of the options | 20:22 |
| kgriffs | let me illustrate... | 20:23 |
| kgriffs | https://etherpad.openstack.org/p/zaqar-scratch | 20:23 |
| kgriffs | at the top there... | 20:23 |
| kgriffs | I will clear the highlight so you can see me type | 20:23 |
| kgriffs | OK, see Line 60? | 20:23 |
| exploreshaifali | yes | 20:24 |
| kgriffs | ok, so I removed that | 20:24 |
| kgriffs | now at the bottom I get rid of the group and helper function | 20:25 |
| exploreshaifali | kgriffs, hey we have decided to make queue/storage/mongodb/messages | 20:27 |
| kgriffs | oops, fixed | 20:28 |
| exploreshaifali | no need of import also | 20:28 |
| *** sgotliv has quit IRC | 20:28 | |
| kgriffs | why not? | 20:28 |
| exploreshaifali | oh yes needed | 20:29 |
| kgriffs | I guess what I have there assumes COMMON_MONGODB_OPTIONS is hoisted into the package namespace | 20:30 |
| kgriffs | you could also be more explicit | 20:30 |
| kgriffs | anyway, that's one way to do it | 20:31 |
| exploreshaifali | yes :) | 20:31 |
| exploreshaifali | many things are clear now | 20:31 |
| exploreshaifali | kgriffs, Thanks alot!! | 20:32 |
| kgriffs | looks like redis and sqlalchemy options are all the same (for now), but it might not be a bad idea to follow the same pattern so it will be ready if/when options diverge | 20:32 |
| kgriffs | exploreshaifali: sure thing! hope that helps. :) | 20:32 |
| exploreshaifali | yes, we should follow same pattern for redis and sqlalchemy too | 20:33 |
| kgriffs | cool, sounds like a plan! | 20:33 |
| riveter | kgriffs: do you have time to answer more questions? | 20:34 |
| exploreshaifali | so first I should submit a patch describing dir structure? | 20:34 |
| kgriffs | riveter: sure | 20:34 |
| riveter | oh sorry | 20:34 |
| kgriffs | exploreshaifali: I would start with a patch that mostly just moves things around, yes. | 20:34 |
| kgriffs | exploreshaifali: btw, this may be a good time to get rid of the queues | 20:35 |
| kgriffs | dir | 20:35 |
| kgriffs | and move everything up one level | 20:35 |
| kgriffs | let's try to catch flaper87 and see what he thinks about that | 20:35 |
| exploreshaifali | yes | 20:35 |
| kgriffs | riveter: what's up? | 20:35 |
| riveter | I'm looking for a good first patch | 20:36 |
| exploreshaifali | but I think moving above queue may won't work because queue dir also contain other files and dirs like transport | 20:36 |
| *** openstackgerrit has quit IRC | 20:36 | |
| riveter | I posted some questions on #1351462, which I notice you first reported | 20:36 |
| kgriffs | exploreshaifali: ah, true. that should be in its own patch. | 20:37 |
| kgriffs | exploreshaifali: don't worry about that then for now | 20:37 |
| exploreshaifali | all right :) | 20:37 |
| exploreshaifali | Thanks! | 20:37 |
| kgriffs | riveter: looking | 20:37 |
| riveter | it's the one about removing duplication from the pool metacontrollers | 20:38 |
| kgriffs | exploreshaifali: thanks for contributing! | 20:38 |
| exploreshaifali | I am enjoying alot | 20:38 |
| riveter | I'm willing to take a crack at making __getattr__ work but I think the methods involved are different enough that it will be pretty ugly | 20:39 |
| kgriffs | exploreshaifali: rock on | 20:39 |
| exploreshaifali | :d | 20:39 |
| exploreshaifali | :D | 20:39 |
| kgriffs | riveter: yeah, so that may be why it ended up the way it is now | 20:39 |
| riveter | so, is it ok to look at other ways to remove some of the duplication? | 20:39 |
| kgriffs | i just feel like if we challenge ourselves we should be able to improve it at least a little | 20:40 |
| kgriffs | riveter: yeah, I think anything is fair game | 20:40 |
| riveter | and if so, do you have any preference for how? | 20:40 |
| riveter | I was thinking just to add a utility method that takes a return-on-fail value | 20:41 |
| kgriffs | riveter: no real preference, other than the less overhead the proxying causes, the better. A few microseconds can have a significant impact on throughput | 20:41 |
| riveter | hm, that's what I was wondering about | 20:41 |
| riveter | the cached getattr is nice for that | 20:41 |
| riveter | so a decorator maybe? | 20:42 |
| kgriffs | perhaps | 20:43 |
| kgriffs | I'd say, start simple and then we can optimize if needed | 20:43 |
| riveter | ok, fair enough | 20:43 |
| riveter | thanks, this is the friendliest irc channel I've seen in a long while | 20:43 |
| kgriffs | heh, glad to hear it! | 20:44 |
| riveter | oh one more thing - I assume the method signatures can't be changed? | 20:45 |
| notmyname | kgriffs: your mother was a hamster and your father smelt of elderberries! | 20:51 |
| kgriffs | LOOOOOL | 20:51 |
| riveter | notmyname: bonjour | 20:51 |
| notmyname | riveter: ;-) | 20:51 |
| *** aj9 has joined #openstack-zaqar | 20:56 | |
| *** jchai_afk is now known as jchai | 20:56 | |
| kgriffs | riveter: hmm, actually, changing the method signatures may be OK. My main concern would be breaking 3rd-party drivers if the contract changes, but I don't think anyone has created any yet. | 20:57 |
| riveter | oh interesting | 20:57 |
| riveter | yeah, cause some things would be easier if an arg could become a kwarg | 20:57 |
| kgriffs | oic | 20:57 |
| riveter | well, not easier, but prettier | 20:57 |
| *** aj9 has left #openstack-zaqar | 20:59 | |
| riveter | anyway I have to go, but I work on this this weekend | 20:59 |
| riveter | thanks again for your help | 20:59 |
| riveter | cheers :) | 20:59 |
| *** cpallares has quit IRC | 20:59 | |
| kgriffs | sure thing, thanks! | 21:02 |
| *** sriram1 has quit IRC | 21:23 | |
| *** sgotliv has joined #openstack-zaqar | 21:34 | |
| *** jchai has quit IRC | 21:40 | |
| *** X019 has quit IRC | 21:46 | |
| *** sgotliv has quit IRC | 21:58 | |
| *** X019 has joined #openstack-zaqar | 21:58 | |
| *** mpanetta has quit IRC | 21:59 | |
| *** earnThis has quit IRC | 22:00 | |
| *** amitgandhinz has quit IRC | 22:15 | |
| *** sgotliv has joined #openstack-zaqar | 22:39 | |
| *** sgotliv has quit IRC | 22:43 | |
| *** openstackgerrit has joined #openstack-zaqar | 23:02 | |
| *** sgotliv has joined #openstack-zaqar | 23:02 | |
| exploreshaifali | kgriffs, around? | 23:09 |
| exploreshaifali | I want to know how transport and storage layers communicate with each other | 23:10 |
| *** sgotliv has quit IRC | 23:32 | |
| *** yfujioka has joined #openstack-zaqar | 23:37 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!