majodev

Wow, Thanks you for this paste!

I'll look into that and maybe provide some additional fonts on a per OS/Browser basis to download. The main problem is that it's impossible to have a static CSS snippet AND provide different files per User-Agent (a javascript solution might help). So until I've found a solution to this, you are absolutely right, these optimizations will get lost on local use!

For reference: I'm currently using the following User-Agents for fetching the font file urls from Google's CSS: http://pastebin.com/5SZ1Q6kj

AFAIK there aren't any .woff, .woff2, .svg, .eot in there + no CSS snippets to use them. Also 2.5GB is quite a big download.

THX for this response! I'll look into that.

Interestingly, Google's own Fonts API which I'm using with my service, doesn't provide any information about the underlying license of each font (I previously wanted to include that info in the UI as well).

Cool, I did not know that .woff is supported in every major browser. I will include a .woff only CSS snippet in the service ASAP.

EDIT: Done.

Based on your User-Agent, Google's CSS service only responds with a CSS to either a .woff, .woff2 .eot, .svg or .ttf font file. For local hosting you'll need to order these formats in a compatibility sequence, so every browser gets the one it supports (nice writeup here: http://css-tricks.com/snippets/css/using-font-face/).

As all files are the exact same ones Google is hosting in the wild, I don't feel like you'll lose any optimizations (which might be the case when using font generation services like FontSquirrel).

Hm, well then I will have to request the font files fetched from their hosted CSS, compress them and pipe this output (the resulting zip) without any caching on the server.

This may result in a higher latency, but might be more safe in regards to their ToS.

I'm really not sure. However, I'm piping are the exact same fonts Google is serving in their hosted CSS-service - in a similar fashion like Clemens Lang's Bash script (https://neverpanic.de/blog/2014/03/19/downloading-google-web...) is doing that.

The only thing that bothers me is the generation of the zip file, but streaming these files without any caching on the server would be very expensive (especially on a single free heroku dynamo).

Great idea.

I recently wrapped Cryptocat with node-webkit (still awaiting PR approval https://github.com/cryptocat/cryptocat/pull/694), hence allow me to give you some lib recommendations:

- Use node-webkit-builder (https://github.com/mllrsohn/grunt-node-webkit-builder) for your build-process (your Gruntfile.js looks very bloated currently).

- Use node-notifier (https://github.com/mikaelbr/node-notifier) for desktop notifications (looks like the best solution so far).

Anyways, best luck to you.

  Location: Austria  

  Remote: Yes  

  Willing to relocate: Yes (within EU, else depends)  

  Technologies: Trained user of Java, T/PL-SQL and ActionScript 3; Proficient in JavaScript (Browser/Node.js), Bash, VB, PHP, HTML and CSS; Expert knowledge in object-oriented programming, build, deployment and testing environments  

  Résumé/CV: http://ranf.tl/cv/  

  Email: see resume (crawl prevent)  

A dedicated Austrian IT professional with considerable experience in a range of software development environments and a specialization in web technologies. Passionate about crafting maintainable software systems to support business processes and creatively solving challenging problems. A proactive team member with excellent communication, planning and research skills, with a distinctive awareness and understanding of the languages of business and IT.