Add Google PageSpeed Apache module

[indolering]

There are three ways to optimize a website:

1. Manually using hardcoded links (i.e. library.min.js).
2. Batch processing using a /dist folder (i.e. grunt generating concat.min.js).
3. Automatically using the server.

Number three is the obvious choice. The easiest way to accomplish this is using Cloudflare but since everyone is super paranoid we can use Google's PageSpeed module.

I would suggest we point namecoin.info to the raw, unoptimized page and keep namecoin.org as the optimized page. Then we don't have to fiddle with PageSpeed to rule-out optimization bugs.

[phelixbtc]

For me the server is super fast...

[ArlasJ]

Sorry,

but I don't know why "Manually using hardcoded links" should optimize the server, because of his speed. It would be good for the SEO but for the speed it hasn't a matter.

Furthermore I prefer to use nginx instead of apache. Nginx shows often that it has 5x more speed than apache.

[indolering]

No, no, this has to do with optimizations at the page level and workflow. When you are developing a website, you use the non-minified JS, HTML, and CSS but when you deploy you want to cut the fat.

It's like writing in machine code vs. an optimizing compiler vs. a JIT. A batch optimization step makes development and maintenance easier as you aren't spending your time with micro-optimizations like pushing your JS to the bottom of the page. PageSpeed just applies the optimizations on the fly, removing the need to manage a development and an optimized version of the site,

It doesn't matter if it's NGNIX or Apache, PageSpeed is available for both. I use Apache because I'm familiar with it. If someone wants to volunteer as a sysadmin, I'd be happy to let them handle the NGNIX configuration.

[ArlasJ]

Yes, there are you right.

I was wondering only, because phelix did write that the server is fast.
And yes of course apache has the same Pagespeed like Nginx. (It is a clientside thing). But you should keep in mind that big organisation pages have a lot of visitors in the same time. And there Nginx is a lot of faster than Apache2. I used apache2 before a lot also. But then I saw the benchmark tests and tried to configure Nginx.

It is very hard for the start. But now I can say that some things are better to handle in nginx than in apache. For example you can write domain names splittet by a blank instead of to write a new server block. I really like it!

[phelixbtc]

Turns out Nginx is already set up on the new server. Hopefully we will be able to move the site soon. We can then worry about speed but I don't think we will have to.

[indolering]

PageSpeed should be set and forget. Since no one seems to be in opposition
to PageSpeed I will assume that we will turn it on at some point and set up
my development workflow accordingly (i.e. link to full libs and forget
about a compilation step).
On Sep 11, 2014 12:32 AM, "phelixbtc" [email protected] wrote:

Turns out Nginx is already set up on the new server. Hopefully we will be
able to move the site soon. We can then worry about speed but I don't think
we will have to.

—
Reply to this email directly or view it on GitHub
#57 (comment)
.

[phelixbtc]

Well, IMHO there are more important things to worry about. Also it is better to keep things simple.

[indolering]

Well, IMHO there are more important things to worry about. Also it is better to keep things simple.

This makes development simpler:

It's like writing in machine code vs. an optimizing compiler vs. a JIT. A batch optimization step makes development and maintenance easier as you aren't spending your time with micro-optimizations like pushing your JS to the bottom of the page. PageSpeed just applies the optimizations on the fly, removing the need to manage a development and optimized versions of the site,

Shobute has taken this on, so I will let him decide, but please stop bike-shedding.

[JeremyRand]

@phelixbtc I don't think "there are more important things to worry about" is really a reason to oppose a change. If someone's willing to work on this, it's useful, and it doesn't introduce a security issue, then I have no objection to it being done.

[phelixbtc]

@JeremyRand Well, it causes work for me and others now and potentially even more so in the future. We need to use our limited resources carefully and keep everything as simple as possible.