> I only see fonts (or infos about fonts + fonts) being retrieved from > fonts.googleapis.com, no CSS, True, its the font, not the CSS.
> Although I may agree with this tracking defect, I guess retrieving > them from a CDN speeds up the page load for browsers that already have > retrieved the fonts for another website. My understanding of the theory: - That is only a slight speed up on the first visit. - For all cdr visits the browser makes GET/HEAD and receives a 304 Not Modified. I checked with Firefox DevTools (F12 -> Network Tab): - in my case, the font file is actually delivered every time (200, not 304 response) - the first two font requests "css?family=Exo:500,300" and "css?family=Open+Sans:300,300italic,400,600,600italic" are each taking around 80-90 ms, these requests that take the longest together with "js.l" delivered from picolisp.com, but "js.l" is like 270 times bigger - so in my case, the Google CDN is not speeding up anything, I would guess the lisp webserver might actually be faster, though your mileage might vary depending on your network distance to the picolisp.com server - testing with Firefox DevTools, I see a 404 Not found for "http://picolisp.com/wiki/wiki/css/typhography.css", but I'm not sure where this one is linked in the page Loading Performance: - Nowadays, especially with the client-heavy "stylish" websites, the bottleneck is usually the Rendering, not the network (can be different for mobile network) - I check this with the Firefox PlugIn "app.telemetry Page Speed Monitor", according to that: -- Page fully loaded after 6030ms -- 5633 ms of that is spent in Processing (Rendering), not in network (- yep, I've got fast internet, so your mileage could differ, but I think the overall pattern will most likely be the same) Summary: - Doesn't look like picolisp.com is benefiting from Google CDN. - picolisp.com is not page with much traffic, so CDN is not needed (ok, maybe when we get slashdotted, but then the fonts will not make much a difference) - in my personal experience, picolisp webserver is usually vastly faster compared to other dynamic website stacks (php, python, ruby...), though I don't have real benchmark data to prove this (so might be wrong) Regards, beneroth ----- Original Message ----- From: Christophe Gragnic [mailto:christophegrag...@gmail.com] To: picolisp@software-lab.de Sent: Fri, 25 Mar 2016 10:11:23 +0100 Subject: Re: Redesigned Wiki is Live On Thu, Mar 24, 2016 at 9:49 PM, <andr...@itship.ch> wrote: > Please host the stylesheets on > picolisp.com instead of using tracking CDN.? I only see fonts (or infos about fonts + fonts) being retrieved from fonts.googleapis.com, no CSS, Although I may agree with this tracking defect, I guess retrieving them from a CDN speeds up the page load for browsers that already have retrieved the fonts for another website. I'm no expert, maybe someone here can clear things up? > Elsewhere-link at the bottom of main page > -> why not generate here a link at random from a predefined list? > -> like a list of links of repos/pages of picolisp-projects ? and the > picolisp wikipedia article? Nice idea ! I don't quite like this link pointing abruptly to google.com too, especially for no real interesting reason. All this said, this is great great work. Many thanks !!! chri -- http://profgra.org/lycee/ (site pro) http://delicious.com/profgraorg (liens, favoris) https://twitter.com/profgraorg http://microalg.info (langage de programmation pédagogique) http://expressions.club/ (structure des expressions mathématiques) -- UNSUBSCRIBE: mailto:picolisp@software-lab.de?subjectUnsubscribe