| # |
Dec 4th 2017, 19:57 |
 admad |
thanks, i'll report back after trying out |
| # |
Dec 4th 2017, 19:56 |
inoas |
both 100% page speed |
| # |
Dec 4th 2017, 19:56 |
inoas |
it is mixed HTTP2/HTTP1 optimization |
| # |
Dec 4th 2017, 19:56 |
inoas |
https://gist.github.com/inoas/2ac1467f83c83af3ea9a41afda905a08 |
| # |
Dec 4th 2017, 19:56 |
inoas |
don't complain about the stuff that should indeed be in middleware or controller context @ response/header() ;-) |
| # |
Dec 4th 2017, 19:52 |
admad |
inoas: gimme the 35 lines :slightly_smiling_face: |
| # |
Dec 4th 2017, 19:49 |
inoas |
next stop --- fighting apache cors settings to fetch fonts from static-domain being on domin |
| # |
Dec 4th 2017, 19:47 |
inoas |
I mean... I need 35lines of js to load jquery so that other scripts can be loaded once jquery is loaded so that I can load jquery deferred... sure. |
| # |
Dec 4th 2017, 19:47 |
inoas |
makes most of inlining, sprite maps, ugly js hacks unnecessary |
| # |
Dec 4th 2017, 19:46 |
inoas |
http2 solved a lot of those ugly problems |
| # |
Dec 4th 2017, 19:46 |
inoas |
inline css/js or Filament loadCSS |
| # |
Dec 4th 2017, 19:45 |
inoas |
idea? http2 is on 80%+ devices |
| # |
Dec 4th 2017, 19:45 |
inoas |
if (str_starts_with(env('SERVER_PROTOCOL'), 'HTTP/2') === true) { header(sprintf('Link: <%s>; rel=preload; as=style', $this->Url->assetUrl($criticalCssAsset)), false); ... you can probably make it a lot cleaner but that's the idea |
| # |
Dec 4th 2017, 19:45 |
admad |
in an ideal world sure |
| # |
Dec 4th 2017, 19:44 |
inoas |
and push/preload |
| # |
Dec 4th 2017, 19:44 |
inoas |
just use http2 :) |
| # |
Dec 4th 2017, 19:44 |
inoas |
no |
| # |
Dec 4th 2017, 19:44 |
inoas |
haha |
| # |
Dec 4th 2017, 19:44 |
admad |
show them a stupid spinner :slightly_smiling_face: |
| # |
Dec 4th 2017, 19:43 |
inoas |
should I show you in a few lines how to get rid of flicker ;) |
| # |
Dec 4th 2017, 19:43 |
inoas |
admad you gotta understand the tech ;) |
| # |
Dec 4th 2017, 19:43 |
inoas |
but that doesnt always work |
| # |
Dec 4th 2017, 19:43 |
inoas |
and we cannot do anything about it unless we cache their scripts on our servers somehow |
| # |
Dec 4th 2017, 19:42 |
inoas |
some projects use so much SEM-shit... and we cannot get past say 80 page speed |
| # |
Dec 4th 2017, 19:42 |
inoas |
but the tards somewhere else |
| # |
Dec 4th 2017, 19:42 |
inoas |
admad well thats not the fault of pagespeed |
| # |
Dec 4th 2017, 19:39 |
admad |
..and things like loading fonts and css in <head>. Fuckers you expect users to see flicker of styles being applied and fonts being switched once DOM is ready? |
| # |
Dec 4th 2017, 19:37 |
obinoob |
hmic it took me a while to figure it! It work's perfect! (y) |
| # |
Dec 4th 2017, 19:37 |
admad |
fucking hate pagespeed. for eg. it whines about not high enough cache time for google analytics script itself. |
| # |
Dec 4th 2017, 19:30 |
inoas |
now the last 2 things pingdom complains about (pagespeed is at 100) is the query strings for asset cache busting (cause some http proxies ignore those when determining ressource identity for caching/busting) and well... cookie-less domains for assets |
| # |
Dec 4th 2017, 19:29 |
inoas |
folf fomf whatever |
| # |
Dec 4th 2017, 19:29 |
inoas |
including no above the fold issues with fonts but still no fouf |
| # |
Dec 4th 2017, 19:29 |
inoas |
for now I am happy with the immensive benefits of http2 |
| # |
Dec 4th 2017, 19:29 |
inoas |
i am setting things in close logical proximity but I will have to find a better way at some point in time |
| # |
Dec 4th 2017, 19:28 |
inoas |
(and maybe we should.) |
| # |
Dec 4th 2017, 19:28 |
inoas |
but then we can just remove the whole <head> block of html :) |
| # |
Dec 4th 2017, 19:28 |
inoas |
well in theory I do, yes |
| # |
Dec 4th 2017, 19:28 |
inoas |
I should not but I was not eager to pull out what stylesheets are required out of layout.ctp |
| # |
Dec 4th 2017, 19:28 |
hmic |
you are using a middleware to to that, or at least utilize the response in the controller |
| # |
Dec 4th 2017, 19:27 |
inoas |
well I am ;-) |
| # |
Dec 4th 2017, 19:27 |
hmic |
:P |