14:30:19 #startmeeting Pulp Triage 2020-05-29 14:30:19 #info fao89 has joined triage 14:30:20 Meeting started Fri May 29 14:30:19 2020 UTC. The chair is fao89. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:30:20 Useful Commands: #action #agreed #help #info #idea #link #topic. 14:30:20 The meeting name has been set to 'pulp_triage_2020-05-29' 14:30:20 fao89: fao89 has joined triage 14:30:44 #info dkliban has joined triage 14:30:44 !here 14:30:44 dkliban: dkliban has joined triage 14:30:46 #info ppicka has joined triage 14:30:46 !here 14:30:46 ppicka: ppicka has joined triage 14:30:53 !next 14:30:54 fao89: 6 issues left to triage: 6851, 6837, 6835, 6834, 6828, 6714 14:30:54 #topic https://pulp.plan.io/issues/6851 14:30:55 RM 6851 - hyu - NEW - Regression in regenerate applicability 14:30:56 https://pulp.plan.io/issues/6851 14:31:24 #idea Proposed for #6851: move to rpm 14:31:24 !propose other move to rpm 14:31:24 fao89: Proposed for #6851: move to rpm 14:31:24 #info ttereshc has joined triage 14:31:24 !here 14:31:25 ttereshc: ttereshc has joined triage 14:31:28 !here 14:31:28 #info x9c4 has joined triage 14:31:29 x9c4: x9c4 has joined triage 14:31:51 +1 14:31:57 #agreed move to rpm 14:31:57 !accept 14:31:57 fao89: Current proposal accepted: move to rpm 14:32:03 +1 14:32:43 !next 14:32:44 fao89: 5 issues left to triage: 6837, 6835, 6834, 6828, 6714 14:32:44 #topic https://pulp.plan.io/issues/6837 14:32:45 RM 6837 - ekohl - NEW - Various links in installation instructions show a 404 page 14:32:46 https://pulp.plan.io/issues/6837 14:32:55 #idea Proposed for #6837: accept and add to sprint 14:32:55 !propose other accept and add to sprint 14:32:55 dkliban: Proposed for #6837: accept and add to sprint 14:32:56 #info ggainey has joined triage 14:32:56 !here 14:32:57 ggainey: ggainey has joined triage 14:32:57 #info ipanova has joined triage 14:32:57 !here 14:32:58 ipanova: ipanova has joined triage 14:33:12 +1 14:33:17 #agreed accept and add to sprint 14:33:17 !accept 14:33:17 fao89: Current proposal accepted: accept and add to sprint 14:33:18 fao89: 4 issues left to triage: 6835, 6834, 6828, 6714 14:33:18 #topic https://pulp.plan.io/issues/6835 14:33:19 RM 6835 - dkliban@redhat.com - NEW - .travis/release.py in pulpcore creates the wrong version for the final .dev version commit 14:33:20 https://pulp.plan.io/issues/6835 14:33:44 i think we need to add more information to the wiki page 14:33:50 but this was mostly user error on my part 14:34:14 the problem for me is: where is the right place for documenting it? 14:34:38 i think both wiki and --help helps 14:34:41 "release.py --help" ? 14:34:43 I'll add a note to put it on the wiki then 14:35:12 #info bmbouter has joined triage 14:35:12 !here 14:35:12 bmbouter: bmbouter has joined triage 14:35:16 #idea Proposed for #6835: accept and add to sprint 14:35:16 !propose other accept and add to sprint 14:35:18 fao89: Proposed for #6835: accept and add to sprint 14:35:20 +1 14:35:22 +1 14:35:22 #agreed accept and add to sprint 14:35:22 !accept 14:35:23 fao89: Current proposal accepted: accept and add to sprint 14:35:23 #topic https://pulp.plan.io/issues/6834 14:35:24 fao89: 3 issues left to triage: 6834, 6828, 6714 14:35:25 RM 6834 - mdellweg - NEW - A pulpcore / plugin pr should be able to point to an experimental version of the pulp-ci base image 14:35:26 https://pulp.plan.io/issues/6834 14:35:43 task? 14:35:50 +1 14:35:54 yeah ... but we should add it to the sprint after that 14:36:15 #idea Proposed for #6834: convert to task 14:36:15 !propose other convert to task 14:36:15 dkliban: Proposed for #6834: convert to task 14:36:15 !propose other convert to a task and add to sprint 14:36:20 haha 14:36:20 +1 14:36:23 #idea Proposed for #6834: convert to a task and add to sprint 14:36:23 !propose other convert to a task and add to sprint 14:36:23 fao89: Proposed for #6834: convert to a task and add to sprint 14:36:27 #agreed convert to a task and add to sprint 14:36:27 !accept 14:36:27 fao89: Current proposal accepted: convert to a task and add to sprint 14:36:28 #topic https://pulp.plan.io/issues/6828 14:36:28 fao89: 2 issues left to triage: 6828, 6714 14:36:28 +1 14:36:30 RM 6828 - fao89 - POST - browsable API missing endpoints on /pulp/api/v3/ 14:36:31 https://pulp.plan.io/issues/6828 14:36:44 #idea Proposed for #6828: accept and add to sprint 14:36:44 !propose other accept and add to sprint 14:36:46 +1 14:37:00 it is good to open floor, due the discussion on the PR 14:37:10 +1 14:37:19 #agreed accept and add to sprint 14:37:19 !accept 14:37:19 fao89: Current proposal accepted: accept and add to sprint 14:37:43 pulpbot is weird today 14:37:43 fao89: Error: "is" is not a valid command. 14:37:49 !next 14:37:50 #topic https://pulp.plan.io/issues/6714 14:37:50 fao89: 1 issues left to triage: 6714 14:37:51 RM 6714 - alikins - NEW - drf builtin manage.py 'generateschema' command fails on pulp base viewsets 14:37:52 https://pulp.plan.io/issues/6714 14:38:37 so think we should convert this to a story or task 14:38:45 the request is to generate openapi v3 schema 14:38:57 we currently only support generating openapi v2 schema 14:39:06 and to go away from drf_yasg 14:39:08 ? 14:39:10 yes 14:39:20 +1 to a story 14:39:23 drf_yasg is not being actively developed anymore 14:39:41 i have noticed that they have not done a release in many months and PRs are not getting merged 14:39:44 drf is not mature for it yet 14:40:02 fao89: i agree, but we shuold keep this on our road map 14:40:07 I tried to use it, and I had to do many tweaks 14:40:07 and maybe use a different project 14:40:10 it probably should be an epic 14:40:20 it will likely be many tasks 14:40:25 bmbouter: +1 14:40:30 #idea Proposed for #6714: convert to an epic story 14:40:30 !propose other convert to an epic story 14:40:31 ttereshc: Proposed for #6714: convert to an epic story 14:40:39 #agreed convert to an epic story 14:40:39 !accept 14:40:39 fao89: Current proposal accepted: convert to an epic story 14:40:41 fao89: No issues to triage. 14:40:53 Open floor! 14:40:58 https://hackmd.io/SVCMjpwXTfOMqF2OeyyLRw 14:41:07 redmine spam! 14:41:36 aw man I meant to look into this earlier 14:41:36 https://pulp.plan.io/issues/5878 14:42:12 I'm not sure what we can do without the api for removing notes 14:42:27 it's the majority of spam, imo 14:42:39 the option discussed last time as a temporary alternative was to require account confirmation 14:42:53 ah I see 14:43:01 yeah ... i think we should require that 14:43:12 i tend to agree 14:43:25 https://www.redmine.org/projects/redmine/wiki/Features#User-self-registration-support 14:44:07 account-confirmation is fine, but nowadays is unlikely to have a significant impact 14:44:22 right now it's set to accoutn activation by email 14:44:23 the SEO-optimizer-folks have scripts to monitor email addresses and auto-confirm 14:44:36 the issue is that we don't want to auto-confirm 14:44:37 ah, kk 14:44:41 these are all humans registering 14:44:59 yah 14:45:07 so the only option is administrators approving accounts 14:45:30 if amins have to verify accounts, we need some kind of comment from the user stating why they are opening the account 14:45:39 I agree 14:45:39 aye 14:45:52 and redmine's process won't change so we would have to have email correspondance 14:46:33 the other option is to not use the API but have the automation drive a firefox browser to remove comments 14:46:37 not a great option obviously 14:47:14 i've done things like that before with selinium 14:47:29 you can do it without an actual GUI browser 14:47:40 that's an option for selinium 14:47:45 yup 14:47:50 I also believe it's diable 14:47:55 doable 14:48:45 from a project perspective the cost of developer time (both in cost and opportunity cost) I think makes manual correspondance not really viable 14:48:54 the spammer cost is extremely low 14:48:58 our cost extremely high 14:49:33 you think they will increase the number of registrations? 14:49:59 even the current rate is several per day 14:50:41 as in if all we had to handle were say 3-5 users per day it's a non-negligable economic impact on pulp 14:50:53 and 0 economic impact on spammers 14:51:13 true dat 14:51:22 I understand that 14:51:24 ttereshc: do you know if we can lock their accounts ia the redmine api? 14:51:50 I'm just not sure when we will be able to do the selenium way 14:52:03 let's say we don't do the selenium way as a thought experiment 14:52:06 bmbouter, they said yes, I couldn;t find it but I didn't try hard. 14:52:48 bmbouter, https://www.redmine.org/issues/22434 seems like no 14:53:23 but you can delete one :/ not that useful but an option 14:53:34 I wonder if it deletes their content too 14:53:44 oh I know it doesn't actually 14:53:47 bmbouter: no... marks it as by Anonymous 14:53:50 it makes their content 'Anonymoous' 14:53:52 yeah 14:53:53 I agree 14:54:25 I think selenium is our best option 14:54:43 and to raise urgency again w/ plan.io 14:54:44 i agree 14:54:59 or put a mechanical turk process in place 14:55:03 do we have an algorithm for identifying spam comments? 14:55:24 (btw, i gotta go in 5 mins) 14:55:47 we could train a bayesian classifier easily, we could also flag comments with non-trusted outbound links (they always contain links) 14:56:18 bmbouter: what do you mean by non-trusted? 14:56:31 we could have a whitelist over time 14:56:35 gotcha 14:56:52 do these users ever create new issues? 14:56:56 lmjachky was looking for a neural network task ;) 14:56:56 they do 14:57:08 ok 14:57:09 I was thinking the same thing 14:57:18 bmbouter: lmjachky expressed interesed to work on this and do some machine learning 14:57:31 I'm +1 on him doing it 14:57:32 we should do it! 14:57:34 heh today ttereshc beats me in replies 14:57:35 +1 14:57:36 +1 14:57:45 i am excited about this 14:57:52 am i crazy? 14:57:52 bmbouter: if this can wait until his return we can keep this for him 14:58:04 I can offer to mentor I have done several neural net supervised and unsupervised AI projects (for fun and profit) 14:58:05 it can def wait 14:58:15 yes it can wait until he is available 14:58:15 sweet! 14:58:30 (if you've done em for profit, you're way ahead of most of the AI industry ;) ) 14:58:33 did anyone else want to mentor him on it? 14:58:36 Lubos will be excited as well 14:59:22 I actually created one last year as well to automatically search github issues for CVEs not yet identified based on language used in the ticket (I evne made a presentation on it w/ a formal analysis) 14:59:32 next topic? 14:59:41 almost, when does he return? 14:59:46 I'm trying to figure out when it would start 15:00:08 18 of june 15:00:11 and also if anyone else wants to mentor so I don't jump into something someone else wanted to 15:00:15 gotta go ... see ya all 15:00:20 seeya! 15:00:24 lata 15:00:59 bmbouter: sold, i think :D 15:01:02 lmk if you also want to join, I can offer ot take an AI to setup a kickoff meeting w/ him around the 18th 15:01:27 bmbouter: that would be great 15:01:33 * bmbouter takes the AI 15:01:47 I'll cc the brno folks since you all have worked closely w/ him too just for awareness 15:01:55 AI about AI 15:02:03 ja 15:02:22 I'm ready to move on ty for the extra time for closure 15:02:52 bmbouter: да 15:03:59 next topic: Can we remove the pulp-admin “category” from Redmine? 15:04:00 +1 to remove pulp-admin category 15:04:02 спасибо 15:04:16 +1 15:04:30 I can take care of doing it just after this 15:04:39 this is my weekly redmine data quality improvement campaign 15:04:46 Whats the use of that category (i any)? 15:05:02 originally it was for issues filed relating to the pulp2 CLI 15:05:04 just not needed 15:05:16 but it never got used meaningfully in any process 15:05:19 +1 15:06:07 with the consent of this group I will take the action after this 15:06:10 next topic? 15:06:18 topic: schema view html - https://github.com/pulp/pulpcore/pull/724#issuecomment-634275398 15:06:47 browsable api could hide endpoints 15:07:00 due html comment yeah so I'm concerned about us modiyfing the regex 15:07:17 which is what prevents my +1 to merge (although ty for the work anyway fao89) 15:07:28 https://github.com/pulp/pulpcore/pull/724/files#diff-9bc1ab0842dca699cb7abce66109298dR67 15:07:32 it does fix the problem 15:07:43 but I have a question 15:08:05 why do we have html here at all? 15:08:06 is about your question I wanted to talk: Should we be using HTML code in ours then? 15:08:23 yeah https://github.com/pulp/pulpcore/pull/724#issuecomment-634275398 15:08:31 exactly! I bring it to open floor for getting opinions 15:09:12 fao89: can you link us to an example of where this html is embedeed in the pulp code as an example? 15:09:45 it is possible to change the view to raw yaml or json, it won't be beautiful, but will show all endpoint 15:10:16 that's a solution though, I want to understand why we have this problem and othe rprojects don't 15:10:36 it is only possible to see it on the html source code: https://pulp.plan.io/issues/6828#note-3 15:11:45 but it's coming from somewhere 15:11:51 it comes from URLField 15:11:55 https://github.com/pulp/pulpcore/pull/724#issuecomment-634267585 15:12:20 DRF URLField > django URLValidator 15:13:15 every view which uses a serializer with URLField will comment the html 15:13:29 ok so that validator end up in the schema because it includes the html comment symbol in the schema spec itself 15:13:41 yep 15:13:56 ty for that I did not understand 15:14:25 it is a regex for validating domain on the URLField 15:14:53 yup 15:15:05 https://github.com/pulp/pulp_file/blob/master/pulp_file/app/viewsets.py#L57 <- Is that an example? 15:15:27 x9c4: this is more like the example I was expecting 15:16:02 if it's pulp code like ^ I think the resolution would be to remove those things 15:16:34 FileRemoteViewSet has a little bit more 15:16:51 Line 118 15:17:25 yup 15:17:51 it is not the problem 15:17:52 I was just inspecting the html file you posted and I agree though that the urlfield brings in that character 15:17:59 fao89: yup I agree 15:18:05 but we would run into that next 15:18:18 I'll try to find an example 15:18:35 your html file shows it, the text here https://pulp.plan.io/attachments/529414 15:18:43 grep that for flake8 also does not like html in the doc strings afaik. (if the corresponding packages are installed.) 15:19:49 from what I see only pulp-ansible uses URLField: https://github.com/search?q=org%3Apulp+urlfield&type=Code 15:20:27 these guys are the ones who are injecting html comment: https://github.com/pulp/pulp_ansible/blob/8e7d519f8a824a45d1d96e3af06428dae7cb397a/pulp_ansible/app/serializers.py#L272-L284 15:22:01 fao89: ic you'r saying those lines are actually bringing that regex into the schema 15:22:05 is that right? 15:22:15 yep 15:22:33 URLField from DRF is what is doing it 15:23:00 the way it validates the url is with this regex which contains so if you inspect the html output of pulpcore, pulp_file, pulp_rpm, all plugins except pulp_ansible do you get any I expected you would? I'm not sure tho 15:25:12 yep, only pulp_ansible has so that field comes from DRF or django? or does DRF use django? 15:26:15 oh I see it's importing from DRF 15:26:19 and that uses django underneath 15:26:31 it comes from DRF, but it uses django validator 15:27:00 is there an upstream DRF and/or django issue about this incompatability when those fields are used w/ openapi schema generation? 15:27:37 I don't know, but it is specific for browsable API 15:27:57 and that's a drf feature yes? 15:28:43 I think it is a too specific problem, for seeing it you have to have a browsable API view of a generated openapi schema with view with urlfield 15:29:09 I think the drf browseable aPI is not compatible with modles that use URLField 15:29:17 yep 15:29:24 and if we had a bug filed against drf like that then pulp could add to its plugin docs that we do not support that 15:29:39 and then we could work w/ pulp_ansible to s/UrlField/CharField/ 15:30:18 I don't think we can accept the patch that modifies the validator itself since that's the validator the code is desiring by using URLField to start with 15:30:32 and have it on docs, for other plugins avoid to use URLField 15:30:38 yes exactly 15:30:42 in the pulpcore plugin writer docs 15:30:58 fao89: I can help work out the details after open floor if that's helpful 15:31:12 we're out of time 15:31:21 I'll add some notes on the issue, and maybe file other issues 15:31:24 yep 15:31:27 #endmeeting 15:31:27 !end 15:31:34 thank you bmbouter 15:31:39 #endmeeting 15:31:39 !end 15:31:46 fao89: ty really! great investigation and work 15:32:13 pulpbot does not want to end the open floor 15:32:13 fao89: Error: "does" is not a valid command. 15:32:18 #endmeeting 15:32:18 !end 15:32:31 !friday 15:32:31 ♪ It's Friday, Friday, gotta get down on Friday ♪ 14:20:06 Open floor will start in 10 minutes (no issues for triage) - https://hackmd.io/SVCMjpwXTfOMqF2OeyyLRw 14:30:55 !start 14:30:55 #info fao89 has joined triage 14:30:55 fao89: Error: Can't start another meeting, one is in progress. 14:30:56 fao89: fao89 has joined triage 14:31:09 #endmeeting 14:31:09 !end