[11:56] gary_poster, ping [11:57] Hey Makyo. I just realized that I was changing the card you had just changed--sorrt about that [11:57] gary_poster, That's okay! [11:57] Just trying to distill it down to a sentence: be more efficient and keep forward momentum in exchanges. [11:58] Makyo, cool. From my perspective, positive/encouraging attitude/atmosphere is an element as well, but that's more fuzzy [11:59] gary_poster, yeah, I couldn't figure out how to fit that in and still sound pithy [11:59] :-) [12:00] gary_poster, I think I'm fighting the lp2kanban bot on one of my cards because I missed 'in progress' and hit 'fix committed'. Even though I changed it back, the bot seems to think it belongs in UX Review. I'm trying again today, hopefully it'll stay [12:00] gary_poster, how you feeling today? [12:00] Nope [12:01] Makyo, :-/ bring it up with bac [12:01] hazmat, fever broke sometime last night, but I think I feel it coming back [12:01] :-( [12:01] gary_poster, Alright. Well, for now, that card still needs a review. Should be quick. [12:02] Makyo, ok can look [12:02] Makyo, link so I can be lazy :-D [12:02] gary_poster, i've got multi-vitamins and advil if its of use [12:02] gary_poster, https://codereview.appspot.com/6776052 [12:02] thanks hazmat. I've got advil if I need it [12:02] ack Makyo on it [12:02] I have emergen-C, too, should it be needed. [12:04] I dunno if that helps fever. This seems like what my older son had about two or three weeks ago. It lasted a long time and we finally took him to the dr simply because he wsn't getting better and dr said that there was a weird virus going around. Doesn't produce scary temperatures, but takes a while to leave. [12:04] Ahh, boo [12:05] the school policy is 24 hours without a fever of 100.6 or something before you come back [12:05] I'll probably try to follow that [12:05] don't want to get other people sick [12:06] benji is getting me a thermometer this afternoon on his adventures [12:06] ok reviewing... [12:20] Makyo, I had to read up on appcache. What you did looks good. Did you consider using touch rather than putting the timestamp explicitly in the file? I think that would be simpler but maybe it is insufficient or less attractive for some reason [12:21] I'll approve with that as a question. Either way is ok by me. [12:25] gary_poster, I suppose it also gives us an indicator in the file we can see from the browser if need be, but now that I think about it, that's kind of a weird case :) [12:25] :-) Up to you Makyo. I've sent the appspot review and approving in the MP... [12:25] gary_poster, Thank you :) [12:26] Moving over to more comfortable area, back in af ew. [12:26] np, thank you for the change. [12:26] cool [13:16] Makyo, just realized downside of this approach: every branch/commit will have the appcache included [13:16] because file always changes on install [13:17] Don't know if touch would solve this, but it might. Maybe it is broken in other ways though [13:18] gary_poster, This is true. Research suggests that some servers will not send the manifest unless the file contents are modified, so it's a trade-off. Its up to you all if it's one that we're willing to make, but I feel that, given that we already ignore the [revision details] entry in the code review, also ignoring the manifest should be alright. [13:18] Makyo, I see :-/ [13:19] I figured it was something like that :-) [13:20] Makyo, does the appcache change mean I need to clear the cache explicitly now in dev? I did in order to see a change to index.html [13:20] No, only if changes are made to cached files (currently only svgs) [13:20] ok cool [13:20] Everything else is explicitly not cached. [13:21] I wonder why I had to clear the cache to get index.html up. Maybe it is not handled the same as our js in the server [13:21] ...doesn't look like it [13:21] that is, I don't see special casing [13:22] * Makyo tests [13:26] hm, restarting server and then reloading works [13:32] Make server will update the appcache. [13:32] * Makyo finds: Regardless of whether you include the address of the current page in the manifest, it will be cached. [13:33] So index.html is automatically cached. [13:33] But all resources such as scripts are not. [13:35] ah right [13:36] I saw that too [13:36] but did not make the connection [13:38] Makyo, might be worth an email to group or something about the two issues: why the appcache will be part of all of our commits (unlike the revision details, this will be an annoying source of conflicts unless we can figure out a way to tell bzr to just be quiet and use the newest version...), and how to make index.html reload [13:40] I know we can do bzr resolve --take-other app/assets/manifest.appcache though that is an extra step. I'll read up on it. [13:45] Or shelve before commit. [13:46] Or the appcache target could do the shelve. [13:46] gary_poster, ^^^ [13:52] Makyo, interesting. The last option wouldn't help though, would it? Seems like that would update the file and then immediately hide the change, which isn't what you want. shelving before a commit would work but for this bzr revert would seem more appropriate... Makyo, here's another idea. Instead of modifying in place, we have a manifest.appcache.in. the makefile adds the timestamp to that file and puts copies it to the expected place. We bzr ig [13:52] nore that generated file [13:52] what do you think of that one? [13:53] (so, IOW, manifest.appcache.in never has a timestamp and is only modified by devs; and manifest.appcache is only generated, and ignored by bzr) [13:53] (and has a timestamp) [13:54] gary_poster, I see. Then if we make a change, just copy the .appcache.in to the .appcache, and make will add the timestamp [13:55] ? [13:55] Makyo, or just rerun make appcache: if the Makefile says that .appcache depends on .appcache.in, then it will know to regenerate the .appcache when .appcache.in changes [13:56] * Makyo nod, tests. [13:57] Makyo, actually now that I am in Makefile frame of mind, I suggest that you make a change like this one also, just to be more proper and to help with this kind of thing...one second, making changes for a pastebin [14:01] Makyo, https://pastebin.canonical.com/77359/ . I have no idea if the actual cp/sed approach I sketched there works. My main point is that appcache should be a phony target, pointing to the real file as a target, which points to the .in file. With the changes I gave in that pastebin, the assumption is that the .in file would live in the same directory as the Makefile. Don't have to do that, but it was the easiest thing to do for a sketch [14:01] notice I put appcache in .PHONY [14:01] That is 100% untested :-) [14:03] gary_poster, That's basically what I was working on, just using $(APPCACHE).in [14:03] oh, yeah, that's nicer [14:04] well, as long as it is ok to serve the .in :-) [14:04] I'm sure it is, just slightly messy [14:04] I'd lean towards not serving the .in myself [14:04] That's fine, then [14:06] Seems to work pretty well. [14:06] cool [14:07] So then bzr add manifest.appcache.in; bzr rm app/assets/manifest.appcache; add 'app/assets/manifest.appcache' to .bzrignore ? [14:08] yes, and you can spell the last step as bzr ignore app/assets/manifest.appcach if you want to [14:08] but they are equivalent [14:08] Ah, cool. [14:10] gary_poster, Thanks for the help! [14:12] Makyo, cool, thank you. Another topic: I made a change that is just slightly too big for a self-review, in my own estimation, but maybe if you approve it I can squeeze it in. :-) Pastebin is here: https://pastebin.canonical.com/77360/ . Main question for you is how offended you are by the "required"/"valid" hack I used to get what I wanted without JS [14:12] I don't know how I made it so long before having to do anything with make [14:12] heh [14:12] Maybe I'll still put it in for review, but I'd still like to know what you think [14:14] My make-fu is a lot higher since I joined Canonical. It's a finicky program, but within a certain class of usage I think I can treat it as "simple" :-) [14:16] gary_poster, I think it looks good, personally. I think all of the browsers we're aiming to support will take :valid [14:16] ok cool Makyo thx [14:30] gary_poster, reproposing to get another look. Added an appcache-force so that if someone updates the index.html file but not the manifest, they can just run that and it will do the update [14:38] Gonna grab a walk around the open space before it gets dark. Will resubmit/email after a +1 [15:10] * Makyo returns, colder. [15:57] Makyo, sorry was away from computer. Just approved with two trivial suggestions. [15:57] Where "away" == about three feet away, granted [15:57] But I have the computer muted :-) [15:57] gary_poster, Thanks :) Will do those and submit before we get a bunch of manifest.appcache things. [15:58] good idea