| [10:11:58] | * jeremyr has quit (Quit: Leaving.) |
| [10:46:09] | * jeremyr has joined #aegir |
| [11:18:28] | * jeremyr has quit (Quit: Leaving.) |
| [16:58:47] | * ybabel has joined #aegir |
| [22:04:52] | * jeremyr has joined #aegir |
| [01:20:01] | * ybabel has quit (Ping timeout: 248 seconds) |
| [01:59:54] | * shaneonabike1 has joined #aegir |
| [02:43:44] | * jeremyr has quit (Quit: Leaving.) |
| [02:47:09] | <jonpugh[m]> | helmo42: ergonlogic colan[m] bgm The drupal 8.4 issue inspired me to make a related change: https://www.drupal.org/node/2912492 |
| [02:47:09] | <hefring> | https://www.drupal.org/node/2912492 => Allow hook_hosting_tasks to specify "command" [#2912492] => 2 comments, 1 IRC mention |
| [03:36:27] | * ybabel has joined #aegir |
| [03:38:18] | * ybabel has quit (Client Quit) |
| [03:38:41] | * ybabel has joined #aegir |
| [03:38:53] | * ybabel has quit (Client Quit) |
| [03:42:24] | * memtkmcc has joined #aegir |
| [03:42:32] | * memtkmcc has quit (Client Quit) |
| [03:43:09] | * memtkmcc has joined #aegir |
| [03:45:44] | <helmo42[m]> | @freenode_jonpugh[m]:matrix.org: task_command looks really useful :) |
| [03:46:08] | <jonpugh[m]> | :) just saw your comment |
| [03:47:16] | <jonpugh[m]> | it's listed right at the top of the drush install instructions |
| [03:47:21] | <jonpugh[m]> | so seems official enough |
| [03:49:31] | <jonpugh[m]> | about to try it out |
| [03:55:16] | * shaneonabike1 has quit (Quit: Leaving.) |
| [04:03:06] | <helmo42[m]> | Hi all, ergonlog1c bgm jonpugh colan cweagans gboudrias memtkmcc .. and any one elese ... Scrum time?. |
| [04:03:36] | <ergonlogic[m]> | hi folks |
| [04:05:16] | <helmo42[m]> | Fore reference ... the current issue is https://www.drupal.org/node/2911855 |
| [04:05:18] | <hefring> | https://www.drupal.org/node/2911855 => Drupal 8.4 support [#2911855] => 21 comments, 1 IRC mention |
| [04:05:18] | <jonpugh[m]> | hellow |
| [04:05:18] | * colan waves |
| [04:06:34] | <memtkmcc> | hi all |
| [04:07:27] | <ergonlogic[m]> | so, to summarize: Drupal 8.4 upgraded to Symphony 3, which breaks Drush 8, forcing us to use Drush 9 site-local for Drupal 8.4 sites |
| [04:07:28] | <ergonlogic[m]> | 8.4+ |
| [04:08:11] | <ergonlogic[m]> | am I missing anything so far? |
| [04:08:40] | <ergonlogic[m]> | the drush-launcher appears to just find/call the site-local Drush |
| [04:08:48] | * aegir has joined #aegir |
| [04:09:59] | <jonpugh[m]> | It turns out it's not exactly forcing us, provision does most things outside of drush commands, I was able to install a site... I think its when we invoke "drush @site cr" or "drush @site updb" directly that it would be a problem |
| [04:10:12] | <jonpugh[m]> | but I'm not sure we actually do that other than in hosting_Tasks_extras |
| [04:10:55] | <ergonlogic[m]> | we bootstrap sites on verify and such, though, no? |
| [04:10:56] | <memtkmcc> | as for drush launcher, the big question is, does it work in both directions? So drush 8 will reliably get response from drush 9 site-context? Not sure if it's even possible? |
| [04:11:12] | <ergonlogic[m]> | to pull the schema versions of installed modules? |
| [04:11:35] | <jonpugh[m]> | nothing is preventing that i think |
| [04:12:08] | <ergonlogic[m]> | if we only need the site-local for stand-alone commands, like 'uli' and 'sql-dump', the the launcher should be pretty straight-forward |
| [04:13:10] | <jonpugh[m]> | testing that out now |
| [04:13:27] | <ergonlogic[m]> | right, so does a site-local Drush 9 scan /var/aegir/.drush? |
| [04:14:12] | <jonpugh[m]> | before we dig in, can I commit the Symfony YML patch so we can install sites? |
| [04:14:14] | <jonpugh[m]> | https://www.drupal.org/files/issues/drupal_8_4_support-2911855-6.patch |
| [04:14:31] | <ergonlogic[m]> | sure, that's pretty straight-forward |
| [04:14:41] | <helmo42[m]> | yes |
| [04:14:52] | <ergonlogic[m]> | but we should continue to work in a branch until we have a viable solution |
| [04:14:52] | <jonpugh[m]> | kk |
| [04:15:11] | <helmo42[m]> | launcer mentions site aliases in it's help message ... but I don't see it being handled |
| [04:15:49] | <ergonlogic[m]> | do you guys have a drupal 8.4 w/ Drush9 git repo or something that we're using to test this? |
| [04:16:03] | <ergonlogic[m]> | we should, imo, try to sync our dev/test environments |
| [04:16:22] | <jonpugh[m]> | I've just been using `drush dl drupal-8.4.0-rc2` |
| [04:16:23] | <ergonlogic[m]> | at least, as far as what we're trying to operate on |
| [04:16:36] | <helmo42[m]> | my latest test was also with a drush dl'ed D8 |
| [04:16:36] | <ergonlogic[m]> | does that come with drush9, though |
| [04:16:38] | <ergonlogic[m]> | ? |
| [04:16:44] | <ergonlogic[m]> | I thought it needed composer |
| [04:16:54] | <jonpugh[m]> | I'm not testing that yet |
| [04:17:11] | <jonpugh[m]> | and it does still have composer |
| [04:17:21] | <memtkmcc> | we rely on bootstrapping Drupal from commands like provision-verify, and we freely move between hostmaster and site context behind the scenes on Verify and Migrate etc, so I'm not sure if drush launcher can handle these context switches we are doing, we will see, I guess.. |
| [04:18:49] | * kristofferw has joined #aegir |
| [04:19:08] | <jonpugh[m]> | I agree, I think the drush launcher isn't what we need right now. |
| [04:19:29] | <ergonlogic[m]> | That's my main concern, Adam |
| [04:19:32] | <jonpugh[m]> | > if we only need the site-local for stand-alone commands, like 'uli' and 'sql-dump', the the launcher should be pretty straight-forward |
| [04:20:12] | <memtkmcc> | yep |
| [04:20:21] | <jonpugh[m]> | replace launcher with a call to local drush |
| [04:20:51] | <ergonlogic[m]> | but local drush won't have provision |
| [04:21:37] | <jonpugh[m]> | uli wont need it |
| [04:21:41] | <ergonlogic[m]> | with Drush9, iirc, there's no such thing as global extensions/commands |
| [04:22:06] | <ergonlogic[m]> | but we call (for example) provision-verify in the site context |
| [04:22:49] | * kevineinarsson has joined #aegir |
| [04:23:12] | <colan> | having every site install provision seems crazy. |
| [04:23:19] | <ergonlogic[m]> | I really don't like the idea of relying on platforms to be properly built in order for Aegir to operate on them |
| [04:23:47] | <ergonlogic[m]> | we'd also need to upgrade provision to Drush9, which is probably a pretty heavy lift |
| [04:24:39] | <jonpugh[m]> | I don't think we need any of that right now |
| [04:25:09] | <ergonlogic[m]> | fwiw, Colan and I have been discussing putting more effort behind AegirNG, which would remove our dependence on Drush as our framework |
| [04:25:15] | <jonpugh[m]> | provision directly includes files from drupal core to run install, updates, clear caches... |
| [04:25:21] | <helmo42[m]> | Having a copy of provision in very site is not an option that's maintainable ... it would force us to better look at a clean separation ... Aegir just pulling the strings in Drupal |
| [04:25:26] | <jonpugh[m]> | I am about to test a ULI |
| [04:25:37] | * mutin-s has quit (Read error: Connection reset by peer) |
| [04:25:44] | <colan> | how much was changed api-wise between drush 8 & 9? It sounds like a lot, so that's not good. |
| [04:26:10] | * mutin-s has joined #aegir |
| [04:26:10] | <jonpugh[m]> | exactly, helmo... it's time to separate |
| [04:26:14] | <kristofferw> | One problem they have found with global installation in drush 9 is the namespace problem. We might have fixed it in drush 9. But the problem comes if drush loads library with one version and the site has different version. It will be conflicts. |
| [04:26:15] | * TwoD has joined #aegir |
| [04:27:30] | <kristofferw> | colan, Drush 9 is a rewrite with OO code. No Drupal 6 or 7 support. |
| [04:27:48] | <jonpugh[m]> | exactly what helmo said... it's time to separate |
| [04:28:03] | <jonpugh[m]> | Drush @d8.4 uli still works BTW |
| [04:28:37] | <jonpugh[m]> | sqlc works |
| [04:29:58] | <colan> | kristofferw: so, said another way, drush needs to load iibraries for different versions of the sites its managing, and that's why it's no longer global. makes sense. |
| [04:30:26] | <memtkmcc> | I really doubt we will be able to use stock Drush. Even for Drupal itself (not to mention the Aegir specific stuff) the upgrade from Drupal pre 8.4 to 8.4+ already looks like a classic chicken/egg loop. My first thought after reading that issue I have linked was that we may need to hack Drush to make it work for us in a way we can control. We can't rely on site level Drush, which we don't control, because it |
| [04:30:26] | <memtkmcc> | e issues, moving forward. |
| [04:30:36] | <memtkmcc> | Or, in fact, we may have to just Drop Drush! ;-) |
| [04:30:50] | <jonpugh[m]> | We should drop drush |
| [04:31:01] | <colan> | Seems like that's the direction we need to move in. |
| [04:31:07] | <jonpugh[m]> | I have been saying we should change provision into a symfony CLI for years :D |
| [04:31:21] | <jonpugh[m]> | but anyway, we can all relax as long as we can keep drush 8 globally |
| [04:31:28] | <jonpugh[m]> | https://www.drupal.org/node/2911855#comment-12277933 |
| [04:31:28] | <hefring> | https://www.drupal.org/node/2911855 => Drupal 8.4 support [#2911855] => 25 comments, 2 IRC mentions |
| [04:31:33] | <jonpugh[m]> | all drush commands seem to work so far |
| [04:32:20] | <ergonlogic[m]> | The upstream issue states that Drush8 works with Drupal 8.4 but doesn't support it |
| [04:32:37] | <anarcat> | use rust |
| [04:32:38] | <anarcat> | ;) |
| [04:32:46] | <kristofferw> | Yes. And talking with people here at Drupalcon Vienna. Drush 8 will continue working for Drupal 8.4+. But there is a but.... |
| [04:32:49] | <ergonlogic[m]> | so, it happens to work, but they aren't committing to maintaining that |
| [04:33:05] | <ergonlogic[m]> | anarcat: :p |
| [04:33:05] | <jonpugh[m]> | but what does that mean? |
| [04:33:08] | <jonpugh[m]> | will it stop working in 8.x ? |
| [04:33:13] | <jonpugh[m]> | drush 8.x i mean? |
| [04:33:22] | <anarcat> | drush has always been a liability |
| [04:33:25] | * jonpugh[m] shoo kitty, |
| [04:33:34] | <anarcat> | it's always been very useful, but damn, at what a cost in the long term |
| [04:33:41] | <ergonlogic[m]> | I think we should try to keep limping along with Drush8 for now, and focus on a full re-architecture to just call Drush to install Drupal moving forward |
| [04:33:41] | <jonpugh[m]> | like we don't know :P <3 |
| [04:33:42] | <colan> | it means we shouldn't invest much more in ddrush as it's no longer guaranteed to work |
| [04:34:03] | <ergonlogic[m]> | yes, Drupal 8.something will eventually break Drush8 irreperably |
| [04:34:10] | <jonpugh[m]> | but drush 8.1.x should always work, yes |
| [04:34:11] | <jonpugh[m]> | ? |
| [04:34:34] | <ergonlogic[m]> | for what? |
| [04:34:53] | <anarcat> | is ddrush a thing? like distributed drush? or a typo? :) |
| [04:35:07] | <colan> | typo :) |
| [04:35:07] | <jonpugh[m]> | for drupal 8.4.x |
| [04:35:30] | <ergonlogic[m]> | it should continue to work up to Drupal 8.3, sorta for 8.4, then it's a crap-shoot |
| [04:35:38] | <kristofferw> | I will try to find some drush maintainers tomorrow at Sprint day and query them more. I will post questions/answers in the issue. |
| [04:35:59] | <ergonlogic[m]> | Drush 8 will, in the near future, no longer support Drupal 8 |
| [04:36:23] | <ergonlogic[m]> | at least, that's my reading of the upstream issue |
| [04:36:28] | <jonpugh[m]> | Right, but because Drupal 8 is the one moving forward... |
| [04:36:51] | <ergonlogic[m]> | it should really be called Drupal 9 |
| [04:38:05] | <jonpugh[m]> | If we can get drush maintainers to commit to drush 8.1.x == drupal 8.4.x we will buy ourselves a good amount of time |
| [04:38:11] | <ergonlogic[m]> | probably caused by Crell not being around to point this out... |
| [04:38:24] | * jonpugh[m] rolls eyes |
| [04:38:26] | <anarcat> | ( ddrush does sound like a friggin nightmare - it's already distributed enough ) |
| [04:41:35] | <ergonlogic[m]> | So, I've been slowly working on AegirNG, based on Drupal 8, using a Celery (Python) queue: https://gitlab.com/aegir/aegir |
| [04:41:38] | <jonpugh[m]> | Ok, so do we have anything left to do except to lock ourselves at Drush 8 for the time being? |
| [04:41:56] | <ergonlogic[m]> | the idea is to have a default Ansible-based engine on the back-end |
| [04:42:11] | <anarcat> | ergonlogic[m]: ha! i was about to troll you guys that you should use python instead of drush - i found rust funnier, but there you go |
| [04:42:36] | <ergonlogic[m]> | I'm all for Python on the back-end |
| [04:42:50] | <memtkmcc> | if Drupal core moves forward without even noticing that it breaks the Drush version people have, because somehow many sites still run on Drupal 7 too, it's definitely a sign that old Drush days are over, and we really need to move forward with this stuff, so we are not trapped suddenly |
| [04:43:31] | <ergonlogic[m]> | jonpugh_: what do you mean? |
| [04:43:40] | <kristofferw> | Also the multisite functionallity is not always thought about. |
| [04:44:05] | <kristofferw> | It seems that moste people only think one site (and one server). |
| [04:44:21] | <memtkmcc> | re: > If we can get drush maintainers to commit to drush 8.1.x == drupal 8.4.x we will buy ourselves a good amount of time |
| [04:44:35] | <memtkmcc> | I wouldn't count on this |
| [04:45:07] | <jonpugh[m]> | why not? |
| [04:45:11] | <colan> | good poing re: multisite. yeah, so we need to get rolling on Ansible or whatever so that when things do break irreparably, we'll have another option. |
| [04:45:13] | <jonpugh[m]> | 8.4 is already at RC2, |
| [04:45:13] | <kristofferw> | Yes. I will try to do that tomorrow. |
| [04:45:46] | <jonpugh[m]> | and 8.1 drush has been out for a while, why can't we rely on that at least until drupal 8.5 comes out? |
| [04:45:46] | <memtkmcc> | they are trying to make drush 8 work with core 8.4+, but are already pretty frustrated and angry, it seems |
| [04:45:51] | <colan> | kristofferw: you'll ask them about multisite? wouldn't hurt to know where they're going with this. |
| [04:46:16] | <jonpugh[m]> | memtkmcc: I'm not seeing any problems right now |
| [04:46:20] | <kristofferw> | colan, Ask the to commit to support 8.4 with Drush 8.x |
| [04:46:46] | <kristofferw> | It seems to be working but everything has not been tested yet. |
| [04:47:19] | <jonpugh[m]> | kristofferw: AFter applying and committing your patch everything in aegir seems fine with 8.4 |
| [04:47:39] | <colan> | ah ok. jonpugh_: the problem is that Drupal 8.4- won't be supposed, so drush 8 won't be useful. everything will be moving their drupal sites forward, or they won't be supported anymore. |
| [04:48:03] | <kristofferw> | jonpugh[m], Yes I saw that. And it is great I will do some more testing with migrating site between server with remote import. |
| [04:48:11] | <jonpugh[m]> | congrats to kristofferw BTW :) Welcome our newest contributor: https://www.drupal.org/node/196005/committers |
| [04:48:12] | <hefring> | https://www.drupal.org/node/196005 => Provision => 0 comments, 1 IRC mention |
| [04:48:32] | <colan> | or rather drush 8 won't support stuff in druapl 8.4+ |
| [04:48:34] | <kristofferw> | But are we using Symfony compontent in other parts of provision |
| [04:48:46] | <jonpugh[m]> | 8.5+ |
| [04:48:51] | <jonpugh[m]> | it works fine, see the issue |
| [04:49:11] | <jonpugh[m]> | https://www.drupal.org/node/2911855#comment-12277957 |
| [04:49:13] | <hefring> | https://www.drupal.org/node/2911855 => Drupal 8.4 support [#2911855] => 26 comments, 3 IRC mentions |
| [04:49:15] | * colan is waiting for another troll from anarcat on my latest typo. ;) |
| [04:49:24] | <jonpugh[m]> | all tasks pass successfully |
| [04:49:27] | <colan> | er, yeah, 8.5 sorry. |
| [04:49:44] | <jonpugh[m]> | kristofferw: not in the parts that bootstrap drupal |
| [04:49:57] | <jonpugh[m]> | I don't believe |
| [04:50:02] | <memtkmcc> | jonpugh[m]: I mean, the time between now and core 8.5 is not that long, at least for the amount of work we have to become drush-free |
| [04:50:14] | <kristofferw> | Sorry need to go now. Drupal Trivial night starts in 10 minutes. |
| [04:50:33] | <jonpugh[m]> | Thanks kristofferw !! |
| [04:50:37] | <helmo42[m]> | kristofferw: goodluck |
| [04:50:43] | <colan> | so what is it? 3 months-ish? |
| [04:50:50] | * colan waves to kristofferw |
| [04:50:55] | <kristofferw> | I will need that. :) |
| [04:51:02] | <memtkmcc> | haha |
| [04:51:02] | <jonpugh[m]> | 6 i thought |
| [04:51:08] | * colan checks the release schedule |
| [04:51:21] | <anarcat> | colan: can't think of anythin, sorry :p |
| [04:51:25] | <jonpugh[m]> | hey better than a critical bug :) |
| [04:51:36] | <jonpugh[m]> | Better than 8.4 release date (Which is like a week? |
| [04:52:32] | <colan> | yep, ~6 months: https://www.drupal.org/core/release-cycle-overview |
| [04:52:39] | <jonpugh[m]> | So let's keep talking about this... deprecating drush means deprecating provision, doesn't it |
| [04:53:35] | <jonpugh[m]> | So this patch is the first step in making hostmaster work without provision: https://www.drupal.org/node/2912492 |
| [04:53:35] | <hefring> | https://www.drupal.org/node/2912492 => Allow hook_hosting_tasks to specify "command" [#2912492] => 2 comments, 2 IRC mentions |
| [04:55:41] | * kevineinarsson has quit (Ping timeout: 240 seconds) |
| [04:55:59] | <jonpugh[m]> | $output = provision_backend_invoke($alias, 'provision-' . $task->task_type, $task->args, $task->options, $mode); |
| [04:56:07] | <jonpugh[m]> | That's where the magic happens :) |
| [04:57:38] | <jonpugh[m]> | provision_backend_invoke > drush_invoke_process > drush_backend_invoke_concurrent > _drush_backend_invoke > drush_shell_proc_open > proc_open! |
| [04:57:45] | <jonpugh[m]> | simple right? |
| [04:58:24] | <helmo42[m]> | domething you dream of every day ;) |
| [04:58:25] | <colan> | That's just going to get messier as time goes on. |
| [04:59:00] | <jonpugh[m]> | drush @hm hosting-task is a good stopping point |
| [04:59:09] | <jonpugh[m]> | that drush command should be able to invoke any process. |
| [04:59:26] | <jonpugh[m]> | no reason to send it down the drush_backend_ rabbit hole |
| [05:03:38] | <colan> | ergonlogic is also trying to see what breaks at the moment (details forthcoming - I'm sitting next to him), but anyway, I think it's time to pull the plug on this whole Drush thing. Cut it loose. |
| [05:04:37] | <memtkmcc> | yeah, and 6 months is definitely better than 2 weeks, but it's also much less than, say, 2 years :-) |
| [05:05:07] | <colan> | We can go with these little patches for now to keep stuff running, but I don't see anything else we'd try being sustainable after 6+ months. |
| [05:05:26] | <colan> | memtkmcc: indeed. |
| [05:10:40] | <ergonlogic[m]> | ok, it looks like we're okay for now, based on kristofferw 's patch |
| [05:10:55] | <memtkmcc> | and drush folks responded: https://www.drupal.org/node/2911855#comment-12278005 |
| [05:10:57] | <hefring> | https://www.drupal.org/node/2911855 => Drupal 8.4 support [#2911855] => 30 comments, 4 IRC mentions |
| [05:11:23] | <ergonlogic[m]> | but there's a horizon quickly approaching, after which we probably cannot rely on Drush as our back-end framework |
| [05:12:31] | <jonpugh[m]> | Here's a fork I made of Terra last year, renaming it to "Aegir CLI". Could be a useful starting point. |
| [05:12:41] | <ergonlogic[m]> | so, maybe 1 year out, assuming the Drush maintainers get Drupal 8.5 supported |
| [05:12:50] | <jonpugh[m]> | Symfony CLI apps that write meta data to YML are pretty straightforward to create |
| [05:13:11] | <jonpugh[m]> | https://github.com/jonpugh/aegir-cli |
| [05:13:27] | * fatguylaughing has quit (Ping timeout: 260 seconds) |
| [05:13:56] | <jonpugh[m]> | we'll have to convert all the drush context data to yml file storage (finally) |
| [05:14:18] | * fatguylaughing has joined #aegir |
| [05:14:50] | <ergonlogic[m]> | Colan, Mathieu and I have been discussing building an Ansible-based back-end, based on https://gitlab.com/aegir/aegir. We'll write up an architecture proposal, and post an issue, where we can discuss that, and other options, such as aegir-cli |
| [05:16:49] | <helmo42[m]> | Very good, lets evaluate our options and try to get started on a plan next thursday |
| [05:17:07] | <memtkmcc> | good idea |
| [05:17:35] | <helmo42[m]> | Something completely different to finish up for today ... Have you all seen/signed the http://publiccode.eu/ petition? |
| [05:19:01] | <ergonlogic[m]> | I have now |
| [05:20:43] | <memtkmcc> | helmo42[m]: signed! https://twitter.com/memtkmcc/status/913483432742793216 |
| [05:28:40] | * hefring has joined #aegir |
| [05:48:18] | * fatguylaughing has quit (Quit: fatguylaughing) |
| [05:50:46] | * memtkmcc has quit (Quit: Leaving.) |
| [05:52:24] | * aegir has left #aegir () |
| [08:28:05] | * theMusician has quit (Ping timeout: 240 seconds) |
| [08:45:02] | * kevineinarsson has joined #aegir |
| [09:05:37] | * kevineinarsson has quit (Ping timeout: 260 seconds) |
| [09:05:53] | * kevineinarsson has joined #aegir |
| [09:06:05] | * kevineinarsson has left #aegir () |
