| # |
Dec 4th 2017, 20:18 |
inoas |
another reason to not concat css files with http2 is that while it is a single persistent binary stream that transfers multiple files, the files are cached and requested over that connection by the browser, still, and thus go into the browsers cache as single entities |
| # |
Dec 4th 2017, 20:17 |
inoas |
admad and one more thing if you are still awake |
| # |
Dec 4th 2017, 20:12 |
inoas |
gnite :) |
| # |
Dec 4th 2017, 20:12 |
 admad |
anyway enough ranting, gn |
| # |
Dec 4th 2017, 20:12 |
inoas |
with http1 it is really hard to do right |
| # |
Dec 4th 2017, 20:12 |
inoas |
with http2 it can be done reasonably |
| # |
Dec 4th 2017, 20:11 |
admad |
yeah clients dont know better, they just want green on pagespeed |
| # |
Dec 4th 2017, 20:10 |
inoas |
stupid cors. |
| # |
Dec 4th 2017, 20:10 |
inoas |
the thing is, organic ranking, for projects that matter commercially or political, is king and everyone is afraid of google downranking you because of sucky pagespeed - and bad bounce rates because somehow bad page speed is a sign of slow pages -> high bounce rates |
| # |
Dec 4th 2017, 20:08 |
inoas |
a pro site (heh) |
| # |
Dec 4th 2017, 20:08 |
inoas |
pingdom is a nicer tool than xvii-pagespeed https://tools.pingdom.com/#!/Jmrqj/cakephp.org |
| # |
Dec 4th 2017, 20:07 |
inoas |
google tag manager, analytics, facebook, bing - all got broken SEM ressources |
| # |
Dec 4th 2017, 20:07 |
inoas |
and then all the big players cannot fix their own shit |
| # |
Dec 4th 2017, 20:07 |
inoas |
and on http2 it matters less anyway |
| # |
Dec 4th 2017, 20:07 |
inoas |
and then if you add gzip it is 0.5kb |
| # |
Dec 4th 2017, 20:07 |
inoas |
that is 150kb big and you can save 3kb |
| # |
Dec 4th 2017, 20:07 |
inoas |
for instance minification... on a html document... |
| # |
Dec 4th 2017, 20:06 |
inoas |
don't get me wrong, I don't believe everything they say is important |
| # |
Dec 4th 2017, 20:06 |
admad |
heh, so we are back to where we started, "fuck pagespeed" :P |
| # |
Dec 4th 2017, 20:06 |
inoas |
though it is a customized bs3 with just the grid and responsive utilites and from styles for bootstrapui cakephp3 plugin |
| # |
Dec 4th 2017, 20:06 |
inoas |
because f**k pagespeed :E |
| # |
Dec 4th 2017, 20:06 |
inoas |
and for http2 I inline load them LOL |
| # |
Dec 4th 2017, 20:05 |
inoas |
http2 is one binary channel |
| # |
Dec 4th 2017, 20:05 |
inoas |
no point |
| # |
Dec 4th 2017, 20:04 |
admad |
why aren't you joining css to reduce requests? for e.g. bootstrap.min.css and base.css |
| # |
Dec 4th 2017, 20:04 |
inoas |
caniuse.com/http2 |
| # |
Dec 4th 2017, 20:04 |
inoas |
and if you don't believe 80%+ |
| # |
Dec 4th 2017, 20:01 |
inoas |
in chromes inspector (the only thing chrome is good for) you can see the http2 preload and prefetch headers |
| # |
Dec 4th 2017, 20:00 |
inoas |
afair |
| # |
Dec 4th 2017, 19:59 |
inoas |
and I think it is woff2 only because http2 push is only supported by woff2 clients anyway |
| # |
Dec 4th 2017, 19:59 |
inoas |
for any subsequent http2 request there is none (if the font is in the cache) |
| # |
Dec 4th 2017, 19:59 |
inoas |
that means for the first request (no cache) there is a small flash of loading font |
| # |
Dec 4th 2017, 19:58 |
inoas |
and the fonts are set to prefetch |
| # |
Dec 4th 2017, 19:58 |
inoas |
in the example above the above the fold css does not complain with pagespeed |
| # |
Dec 4th 2017, 19:58 |
inoas |
the browser already knows about your ressources |
| # |
Dec 4th 2017, 19:58 |
inoas |
and then use regular html->css() |
| # |
Dec 4th 2017, 19:58 |
inoas |
but as a PoC you can just try the header commands |
| # |
Dec 4th 2017, 19:57 |
inoas |
<?php if (config('debug') === true) : ?> probably doesnt exist |
| # |
Dec 4th 2017, 19:57 |
inoas |
(.*) https://%{HTTP_HOST}%{REQUEST_URI} [NC,R=301,L] <= thatfor |
| # |
Dec 4th 2017, 19:57 |
inoas |
jfyi |
| # |
Dec 4th 2017, 19:57 |
inoas |
http2 only runs on https |