| === tmpRAOF is now known as RAOF | ||
| === Malsasa is now known as Guest91062 | ||
| === Malsasa_ is now known as Malsasa | ||
| === Malsasa_ is now known as Malsasa | ||
| cimi | greyback_, ping | 09:22 |
|---|---|---|
| greyback_ | cimi: pong | 09:22 |
| cimi | greyback_, hey doc | 09:22 |
| cimi | greyback_, I know you answered already, I forgot :) | 09:22 |
| greyback_ | cimi: 'sup! | 09:22 |
| cimi | greyback_, if I create an Image/BorderImage and I set visible to false, is it loaded by qt but not rendered? | 09:23 |
| cimi | or is not even loaded? | 09:23 |
| cimi | greyback_, adding shadows to cards in dash, wondering if I should put the shadow in a loader since for now is only for application scope | 09:24 |
| greyback_ | cimi: it is loaded by the CPU, and pushed to GPU texture memory | 09:24 |
| greyback_ | but is not rendered | 09:24 |
| greyback_ | cimi: I was chatting with loicm about this | 09:24 |
| cimi | greyback_, so I think I will put it in a Loader | 09:24 |
| cimi | greyback_, yeah, I heard design wants shadows in _many_ places | 09:24 |
| greyback_ | cimi: could you try replacing the BorderImage with 4 images surrounding the application surface? | 09:25 |
| cimi | greyback_, I am doing a different thing | 09:25 |
| greyback_ | cimi: what's your thinking for your approach? | 09:25 |
| cimi | greyback_, I am adding shadows to cards in dash now | 09:25 |
| cimi | greyback_, not app spread | 09:25 |
| greyback_ | cimi: okay. I suggest you don't do anything fancy until you have proved the standard approach is costly | 09:26 |
| greyback_ | shadows are not costly on their own | 09:26 |
| greyback_ | but a bunch of shadows overlapping is costly | 09:26 |
| cimi | greyback_, for the app spread, we want to slice and just use 3 shadows I would say, not sure we need the right edge shadow unless for the rightmost app | 09:26 |
| greyback_ | cimi: agreed | 09:27 |
| greyback_ | but perhaps the code complexity implementing that is worse than that right edge shadow | 09:27 |
| cimi | greyback_, going back to cards in dash... we want shadows under apps in app scope | 09:27 |
| greyback_ | is something to be played with | 09:27 |
| greyback_ | under the app icon? | 09:27 |
| cimi | greyback_, so far I have image & borderimage under ALL cards with visible to false, since you just told me there is CPU loading the png etc etc, I might just put those in a loader that loads only in app scope | 09:29 |
| cimi | yes | 09:29 |
| greyback_ | cimi: why are you thinking of these optimisations? Have you implemented the simple solution and found it extremely slow? | 09:29 |
| cimi | greyback_, no | 09:30 |
| greyback_ | we should only optimize when we identify a problem | 09:30 |
| cimi | greyback_, no | 09:30 |
| greyback_ | yes | 09:30 |
| cimi | greyback_, we should write reasonable code at first | 09:30 |
| greyback_ | reasonable = simple to maintain IMO | 09:30 |
| greyback_ | optimizations introduce complexity | 09:31 |
| cimi | greyback_, we have hundreds of scopes, hundreds of cards, and ALL will load one image & borderimage that will never use | 09:31 |
| greyback_ | complexity should have a good justification | 09:31 |
| cimi | putting the inside a stupid loader won't hurt I think | 09:31 |
| ltinkl | cimi, Image uses caching, so I suppose it won't hurt that much | 09:31 |
| cimi | ltinkl, yeah but they are not used.... | 09:32 |
| greyback_ | cimi: that's your intuition speaking though | 09:32 |
| greyback_ | do you have evidence for this statement? | 09:32 |
| greyback_ | qml does lots of clever things to avoid loading too much | 09:32 |
| ltinkl | cimi, now but chances are the image is already loaded and decoded when you create a new one with the same source | 09:32 |
| greyback_ | we should only optimize when we have to, not because we feel we should | 09:32 |
| ltinkl | agree there, "don't fix what ain't broken" :) | 09:33 |
| ltinkl | cimi, tried to write a simple benchmark with and without a loader? | 09:34 |
| greyback_ | you may end up writing complex code which has no benefit over the simple code | 09:34 |
| greyback_ | I say, write the simple code first. If things end up slow, use a profiling tool to detect the most expensive things and optimize those | 09:34 |
| cimi | ltinkl, but it will load at least once | 09:35 |
| cimi | ltinkl, we can avoid this one | 09:35 |
| greyback_ | hmm, internet wonky today | 09:36 |
| greyback_ | <greyback_> cimi: that's your intuition speaking though | 09:36 |
| greyback_ | <greyback_> do you have evidence for this statement? | 09:36 |
| greyback_ | <greyback_> qml does lots of clever things to avoid loading too much | 09:36 |
| greyback_ | <greyback_> we should only optimize when we have to, not because we feel we should | 09:36 |
| greyback_ | <greyback_> you may end up writing complex code which has no benefit over the simple code | 09:36 |
| greyback_ | <greyback_> I say, write the simple code first. If things end up slow, use a profiling tool to detect the most expensive things and optimize those | 09:36 |
| cimi | greyback_, all right doc | 09:38 |
| === facubatista is now known as Facu | ||
| === alan_g is now known as alan_g|lunch | ||
| === MacSlow is now known as MacSlow|lunch | ||
| ChrisTownsend | popey: Hey, are you around? Wanted to follow up on your unity8-lxc issue. | 12:20 |
| popey | ChrisTownsend: hey! | 12:25 |
| popey | ChrisTownsend: just grabbing a sandwich, what should I be looking for? | 12:25 |
| popey | (in unity, not in my sandwich) | 12:25 |
| ChrisTownsend | popey: lol...So I think the LXC is not starting up for some reason. Let's start from a fresh baseline. Could you run 'sudo unity8-lxc-setup --rebuild-all --redownload"? | 12:26 |
| ChrisTownsend | popey: And let's make sure that completes with no errors. Then we'll go from there. | 12:27 |
| popey | ok | 12:28 |
| popey | ChrisTownsend: http://paste.ubuntu.com/11701731/ that look right to you? | 12:32 |
| ChrisTownsend | popey: Yep, looks good. | 12:33 |
| popey | so reboot and login to unity8? | 12:33 |
| popey | for a nice clean session | 12:33 |
| ChrisTownsend | popey: Yep, let's give it it try. If it doesn't work, then we'll dig some more. | 12:33 |
| popey | ok | 12:33 |
| popey | ChrisTownsend: well! now I get a welcome screen, never had this before! | 12:37 |
| ChrisTownsend | popey: Awesome! It works! | 12:37 |
| popey | my "phone" is now ready to use \o/ | 12:38 |
| popey | thanks ChrisTownsend | 12:38 |
| ChrisTownsend | popey: lol, a big phone like 1995. | 12:39 |
| ChrisTownsend | popey: Sure thing! | 12:39 |
| popey | i suspect it was down to that upstart install | 12:39 |
| ChrisTownsend | popey: Yes, I had to add some workarounds pretty recently in getting upstart into the container. It simply does not like systemd right now. | 12:40 |
| popey | yeah, phone is upstart for user session | 12:40 |
| ChrisTownsend | popey: This is upstart for the system. | 12:40 |
| popey | oh | 12:41 |
| popey | ok | 12:41 |
| ChrisTownsend | popey: systemd just did not play nice with the whole lxc thing I have set up there, so the easiest thing to do was to whack systemd and install upstart for the system services. | 12:41 |
| popey | ah | 12:42 |
| ChrisTownsend | popey: And you probably tried setting up unity8-lxc while I was trying to nail all that down. | 12:42 |
| popey | probably :) | 12:42 |
| ChrisTownsend | popey: And it just wasn't installed correctly. | 12:42 |
| ChrisTownsend | popey: Anyways, I'm glad it works for you now. | 12:42 |
| * greyback_ 's router giving trouble, bbiab | 12:48 | |
| === alan_g|lunch is now known as alan_g | ||
| === MacSlow|lunch is now known as MacSlow | ||
| === alan_g is now known as alan_g|EOW | ||
| === Trevinho is now known as Trevinho|Holiday | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!