[15:33] https://code.launchpad.net/~doismellburning/launchpad/+git/launchpad/+merge/385644 - my og_title assert is failing and it's not yet clear to me if my template code is wrong or I should be poking makeTemplateView to inject more info - would appreciate feedback please [15:33] (both specific and general!) [15:34] SpecialK|Canon: What's the error message? [15:35] cjwatson: `testtools.matchers._impl.MismatchError: 'http' not in None` [15:35] presumably because `request/URL` is None [15:35] I normally write more like content.find('meta', property='og.title'), but your syntax should work too [15:35] request/URL is used in og:url rather than og:title though [15:35] cjwatson: property is a builtin in some modern Pythons though [15:35] cjwatson: hence I went for the dict [15:36] Ah yeah, fair point [15:36] (otherwise I would have) [15:36] "modern" :) [15:36] bah sorry, og_url not og_title [15:37] https://pastebin.canonical.com/p/y7S2VGN6YQ/ in full [15:37] https://pastebin.ubuntu.com/p/DFWsYspMcf/ even [15:38] very few things declare recommended_canonical_url, fine (I do mean to add tests for some of those later) [15:38] Mkay, having a look [15:38] I think soupmatchers would make this clearer FWIW [15:39] I always like composite tests to test as much as they can at once, because it means you get better error messages [15:43] Hm that does look interesting [15:44] request/URL should evaluate to http://launchpad.test here, since that's what makeTemplateView sets up ... [15:45] That's what I assumed; I'm struggling to find much in the way of docs about `request` though [15:45] * cjwatson sets a breakpoint on recommended_canonical_url [15:45] It's a LaunchpadTestRequest in this case [15:46] https://zope.readthedocs.io/en/latest/zopebook/AppendixC.html#tales-overview helpfully tells me it's a thing, but not much more [15:46] Ah [15:46] ...yes in hindsight it is, isn't it [15:47] Which is a thin layer over https://zopepublisher.readthedocs.io/en/latest/browser_api.html#zope.publisher.browser.TestRequest [15:50] As for template language docs, I generally point at https://pagetemplates.readthedocs.io/en/latest/ instead of that old Zope 2 doc [15:51] Digging a little more I suspect I may instead want view/url or context/url based on other usages, but I've not yet found anything definitive [15:51] Ah thanks [15:52] Ah [15:53] SpecialK|Canon: alternate expressions after | only get evaluated if the previous expressions *fail*, not just return None [15:53] Well, https://pagetemplates.readthedocs.io/en/latest/history/TutorialPart2.html?highlight=view%2Furl#inserting-text uses request/URL, but I'm still not clear on the exact different semantics [15:53] SpecialK|Canon: Try "python: view.recommended_canonical_url or request.URL" instead [15:53] cjwatson: ...that'll do it, thanks [15:55] hurrah, cheers :) [16:13] hopefully make css generation repeatable enough: https://code.launchpad.net/~twom/launchpad/+git/launchpad/+merge/385678 [16:13] note, this disables compressed output for now, it makes the file ~80K bigger, but it's a single file that is then cached [16:13] but means SASS leaves the comments in, so I can work out which file is which! [16:14] I'll re-enable compressed output when this bug is confirmed fixed [16:15] Will you be able to compare between qastaging/staging/dogfood or something? [16:15] r=me anyway [16:15] that's the hope, yes [16:16] we have enough environments that build this that I can see if we get consistent results from it [16:16] I kind of dislike that we have rules that are dependant on the ordering of the filesystem [16:16] (apparently) [16:19] Yes, we clearly shouldn't [16:21] hopefully this will fix it for now, a better fix would probably be to create a SCSS _index file and use that rather than the jsbuild mechanisms [16:23] Yeah [16:23] Bah, another buildbot failure due to me [16:33] https://code.launchpad.net/~cjwatson/launchpad/+git/launchpad/+merge/385680 - quick testfix review? [16:37] looking [16:38] cjwatson: +1 [16:38] cheers [16:45] I think that's about all the useful stuff I'm going to get done today. Have a good weekend [16:47] cjwatson: you too! [16:47] https://code.launchpad.net/~doismellburning/launchpad/+git/launchpad/+merge/385644 if anyone's still around...? [16:50] mind blanked, what's the env var I need to use lp-shell with a local instance? [16:50] disables the SSL validation [16:52] LP_DISABLE_SSL_CERTIFICATE_VALIDATION= [16:52] that would do it [16:56] so https://dev.launchpad.net/Running - presumably "Point your usual web browser at https://launchpad.test, and accept the local self-signed certificate." predates "it is advisable to try Launchpad in an LXC/LXD container" [16:57] Probably, but I don't get the connection [16:58] is there anything more modern somewhere else? [16:58] What is your objection? [16:58] cjwatson: well the latter means that doing the former won't achieve much [16:58] Not so [16:58] cjwatson: because I'm running LP in a container [16:58] Works fine [16:59] You do have to follow the link to https://dev.launchpad.net/Running/RemoteAccess [16:59] and my host is not aware of launchpad.test [16:59] Right [16:59] It will be if you do the stuff in Running/RemoteAccess :) [16:59] Running may just need to be reordered a bit [17:00] "Works fine" *if* you have read some as-yet-unmentioned docs is...a stretch ;P [17:00] Yeah, I hadn't noticed that RemoteAccess is actually mentioned further down the page [17:00] But yes, thanks for the pointer, I'll rejig that a bit [17:00] A few people have been through the docs without mentioning that :) [17:01] * tomwardill always reads the entire paper before answering questions ;) [17:01] https://code.launchpad.net/~twom/launchpad/+git/launchpad/+merge/385683 add gitrepository.id to the API [17:01] Good catch that it's misordered though [17:01] (one of my previous statements is a complete lie) [17:02] I think at least myself and pappacena were given both links at the same time, so it's less obvious [17:02] Could be [17:03] https://dev.launchpad.net/Running/LXD also emphasises it a bit at the end [17:03] Anyway, not here :) [17:03] * tomwardill -> also not here, before dog eats anything else he's not meant to because I haven't fed him yet [17:06] Cheers :) [17:06] Have good weekends!