13:02:15 #startmeeting 13:02:15 Meeting started Wed Aug 13 13:02:15 2014 UTC. The chair is alinefm. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:02:15 Useful Commands: #action #agreed #help #info #idea #link #topic. 13:02:15 #meetingname scrum 13:02:15 The meeting name has been set to 'scrum' 13:02:28 #info Agenda 1) Status 2) Open discussion 13:02:28 anything else? 13:04:45 #topic Status 13:04:45 #info Please provide your status using the #info command: #info 13:05:06 #info royce bebased and merged image based template 13:05:24 #info royce fixed wrong bus type in power, refactored device name generation 13:05:40 #info royce helped to address problem of cdrom update, sent 3 patches 13:06:04 #info pvital submitted a patch to change the default environment configuration to production mode. 13:06:23 #info alinefm updated makefile to include spice.css 13:06:53 #info alinefm fixed kimchi running on power by removing a hard coded x86_64 arch 13:06:59 #info pvital is working in a patch set to make upstream Kimchi fully enabled on Power machines 13:07:09 #info pvital reviewed some patches 13:07:10 #info alinefm fixed issue #340 13:07:33 #info alinefm sent patch to properly list host partitions for Ubuntu 14.04 server 13:08:01 #info alinefm sent patch to request /config/capabilities just once on UI and as soon as possible 13:08:14 #info vianac reviewed patches on the mailing list 13:08:25 #info alinefm is working with royce and christy to fix problems while updating cdrom value 13:09:44 #info rotru working to enable kimchi open HELP pages from plugins 13:11:24 anything else? 13:12:41 ok... 13:12:51 today we are finish the sprint 1 stabilization 13:13:06 which means we need to start working on the sprint 2 features now =) 13:13:27 https://github.com/kimchi-project/kimchi/wiki/Todo-1.3 13:14:00 there are just 2 tasks without owners at the moment 13:14:21 anyone would like to pick one of them? 13:15:24 I can do the UI work of the Authorization 13:15:53 sure, wenwang! 13:15:57 I will update the wiki 13:16:46 more volunteers? =) 13:17:01 Also, I think I can do the work of listing Kimchi peers 13:17:17 I will handle list kimchi peers 13:17:29 wenwang, that would be great! but make sure to don't overload you 13:19:34 I will set the list kimchi peers to YuXin as the pci passthrough UI is already done 13:20:14 wenwang, about the authorization one: I talked to YuXin to make the full "edit" dialog available when the vm is running and disable what does not make sense for a running vm 13:20:21 you can talk to him for more details too 13:20:47 alinefm: sure 13:21:26 Display iSCSI targets when iSCSI server is provided (backend done in 3c6c5c) 13:21:40 seems like this one also has no owner 13:22:26 aline, assign this to me also 13:22:27 YuXin, do you think you can hold this one too? 13:22:33 ok 13:24:13 As "PCI Passthrough" is done actually, so I can handle "list kimchi peers" and "iSCSI targets" 13:24:27 then each of UI member has 2 items 13:24:35 sounds a good plan 13:25:01 about the non-assigned backend tasks I will check if royce can pick one and the upload task I need to check with sdxaio 13:25:49 alinefm; you can set "Create template from VM" to me 13:26:06 oh! thanks, rotru! 13:27:10 so I want to see at the ML by next Monday (08/18) a RFC, a V1 patch, a mockup UI or something 13:27:29 we need to start the feature discussion as soon as possible 13:27:46 sounds great! 13:27:54 rotru, shaohef, vianac, YuXin, ok? ^ 13:28:12 alinefm; ok! 13:28:20 wenwang, YuXin, the mockup UI does not need to have a code! it can be just an image 13:28:42 alinefm, ok 13:28:48 alinefm: Okay, np 13:30:08 #action everyone assigned to a task will send to ML a RFC, a V1 patch, a mockup UI or something by next Monday (08/18) 13:30:26 any question? or we can move on... 13:31:15 aline, for UI mockup, we need to backend RFC to have the API designed first 13:32:19 or at least some one send out a description of the feature about what content and information need to be presented 13:33:27 right! I can send an overview for all tasks by EOD tomorrow 13:33:34 many thanks 13:33:47 thanks alinefm 13:34:01 and then each owner can make suggestion and propose changes on it 13:34:06 alinefm: OK 13:34:09 exactly 13:34:11 that way we don't block the UI 13:34:18 exactly 13:34:55 #action alinefm send an overview for all tasks by EOD tomorrow (08/14) 13:35:36 if API can be designed first with url and json response, it will greatly save UI from blocking 13:35:51 Aline, I believe you will shape a good model for everyone 13:36:55 I hope too =) 13:37:03 let's move on to open discussion section now 13:37:38 #topic Open Discussion 13:37:47 what do you want to discuss today? 13:38:14 from shaohe's VNC password patch, I see something to discuss 13:38:43 please, go ahead 13:38:44 there are already many options in VM edit, and it will ever evolve in the furthre 13:38:58 I have a topic 13:39:19 now, we save at an unit per tab 13:39:40 but I think it is better to save what a user truly change 13:40:24 from virt-manager, I see for each item like cpu, memory, if any item changed, it need to be save at once 13:40:24 are you saying even if I don't change any value, all fields are send to the PUT request? 13:41:02 from the design, I think it is saved per tab, right? 13:41:11 right 13:42:02 {"name":"VM-01","cpus":2,"memory":1024} 13:42:15 I only changed cpu, but all are posted to server 13:43:14 now, shaohe add vnc password as a separate tab, aline, you have comment to add it also to general tab 13:44:14 yes 13:44:29 I proposed that as it is part of the same PUT request made on general tab 13:44:37 if user only update cpu, others should not be posted and be updated on vm also 13:44:44 agree 13:45:07 what about use the same design of network and storage tab? 13:45:17 that is what I prefer 13:45:20 the first view, we display only labels 13:45:28 with an edit icon 13:45:33 aline, this is about granularity 13:45:42 when selecting edit, this icon turn to save one 13:45:55 and on each click on save we perform a request 13:45:57 so for each item, we should give user a way to save it separately, right? 13:46:06 yes 13:46:20 that is what we did for storage and network, right? 13:46:30 yes 13:46:44 and we also support save multiple items in batch 13:46:56 so I think here we can copy virt-manager 13:47:08 virt-manager use a button named 'apply' 13:47:36 so for each item in general tab, we follow an 'apply' button 13:48:02 if user changed that item, enable the apply button, and if user click on that button, post to save it 13:48:13 but the dialog window does not close 13:48:29 sounds good for me 13:48:35 but can't we use the same edit/save icons used on network/storage tabs? 13:48:39 if user click on OK button, post all changed field to save them and close the dialog 13:48:42 instead of this apply button 13:48:59 that is ok 13:49:23 I will try to get a mockup out for comments 13:49:32 I have worked in a problem relates to plugins and their help page. 13:49:33 Currently, kimchi is not able to open a help from a plugin, because of the tabe url link (plugins have #plugins/XXX instead of #tab/XXX). 13:49:33 This is easy to fix, however, we need to set a static dir to all plugins help page. 13:49:33 We have some options: 13:49:33 1) Force plugins to copy help pages to Kimchi's ui/pages/help path (I don´t like this) 13:49:34 2) Force plugins to have the same structure that kimchi: plugins/XXXX/ui/pages/help/LANG/ 13:49:36 Choosing (2), Kimchi could set the static path to each plugin help page on the fly when it loads them. 13:49:40 What do you guys think ? 13:49:56 YuXin, that would be great! but anyway, I think it is a good approach 13:50:09 ok 13:50:10 YuXin, are you be responsible to make those changes? may I add a task for it? 13:50:29 ok 13:51:27 moving on to help problem pointed by rotru 13:52:27 1) I don't like it. Each plugin should have its own dir structure 13:52:59 2) For this option, we need to make it configurable IMO. One plugin may not have a help page 13:53:04 how handle this case? 13:53:20 or we will force each plugin to have a help page (which I don't like) ? 13:54:09 alinefm; Its possible to not force 13:54:10 shaohef, vianac ^ 13:54:46 alinefm; in this case, kimchi would fails with "URL not found" .. .then move to guest tab ( this is what happens today ) 13:55:41 alinefm; or, we can create a default kimchi html, saying that help is not available for that plugin/tab 13:55:49 ouch! 13:56:07 rotru, yeap! this is what I'd say 13:56:08 in UI, whether each plugin tab has a ? to lauch its help content, or plugin help content is merged to kimchi help and all lauched with that ? of kimchi? 13:56:46 each plugin has its own help 13:56:54 it is a separated html file 13:57:13 sorry, each tab has its own help (as a plugin can have multiple tabs) 13:57:28 then there will be a ? along with each plugin tab? 13:57:37 yes 13:57:45 it is part of the framework 13:58:07 as the ? appears under the logged user menu 13:58:34 rotru, but cherrypy starts up without errors, even point to it a non-existent dir? 13:58:37 plugin has a xml configuration file, add an entry in that xml file for each tab with the help.html 13:59:11 then we click the ? in each tab to lauch the tab help can get it to work? 13:59:17 YuXin, oh! that will solve all problems =) 13:59:18 rotru, ^ 13:59:22 alinefm; I see some cherrypy errors in the console, but did not pay attention at this moment ... 13:59:53 YuXin, and if the tab does not have a help page setup on tabs.xml we remove the ? option from the menu? 14:00:03 yes 14:00:03 YuXin; I can check this option 14:00:41 YuXin: do you mean put the help.html path in xml configure file? if no help.html path , then a default help.html? 14:01:14 without help.html, does not show the ? on UI, so user can not launch it 14:02:18 I worry about the UI design, how to make it looks good and from user's perspective, plugin and kinchi base are integrated well 14:02:24 YuXin; we cannot add the full help.html path in the xml, because the path is defined on the fly 14:03:02 rotru, in the xml you defined the API to access the file 14:03:06 YuXin; oh, nevermind what I said 14:03:06 not the real path on system 14:03:21 alinefm; I see , just open the xml 14:03:25 YuXin: can a API for help.html? GET /tabs/helps ? 14:03:52 shaohef, we already have it 14:03:53 shaohef; plugins does not use 'tabs' 14:03:56 but the API is /help 14:04:11 shaohe, the tab.xml will pass to client, and just a entry there will get all the configure and UI processing will be straight-forward 14:04:38 I think the main problem is on backend 14:04:53 how to expose or not the plugin help dir? 14:05:03 host guest template storage pool network administration(?) xxxx(?) 14:05:19 the best way to do that would have the plugin config dir has a property for it 14:05:45 rotru, in this case you will need to have a plugin.conf.in and substitute the values on build time 14:06:11 BUT for it works for install and local tests, the plugin must be set as a git sub-module 14:06:25 I mean, a git sub-module on local git repo 14:08:41 http://fpaste.org/125249/14079388/ 14:09:05 we already pass all the plugin config to cherrypy 14:10:17 alinefm; there is not guarantee that /foo/bar/ path will be always used 14:10:58 I know, because that you need to build it on build time using autoconf 14:11:02 like we did on kimchi 14:11:16 @datadir@/ui/help 14:11:27 and substitute it for the real path on build time 14:13:20 alinefm; I see 14:13:27 This is an option too 14:13:59 is there any special reason to you don't use git sub-module on local setup? 14:14:41 alinefm; I am not sure about this 14:17:18 rotru, could you do a test on it? 14:17:30 alinefm; sure 14:17:47 I mean, generate your help dir config on build time + git sub-module and see it if works for local setup? 14:18:01 and the test the installation too, of course 14:18:27 *and then 14:19:04 alinefm; ok 14:19:34 rotru, and let me know if you have any special reason to don't use git sub-module 14:20:12 zhshzhou; ^? any idea ? 14:22:33 seems we have a good plan 14:22:46 we are over time team 14:23:00 feel free to send to ML any other topic you want to discuss 14:23:35 thanks everyone for joining and for the great discussions too 14:23:57 #endmeeting