Nodebb design

[Panda]

When I first thought on the topic of how I would design a forum, I envisaged a mostly JS front-end app, using the back end to access the Database.
That has some advantages and disadvantages. One negative is not being so good for SEO as more Server side rendered forums, if web crawlers dont run the JS to read the forum.

How Nodebb builds its pages is different than I might have guessed. Its heavily reliant on Templating and CSS.

Anyway before getting to that, I have been looking at how Nodebb builds its pages.

On Nodebb forum one of the devs said
The “read api” is used by the NodeBB frontend when loading pages
The “write api” is preferred for server-to-server interactions.

I asked for clarification on what server-to-server interactions might be?
That part of my question was missed, can anyone add some explanation to help me understand?

[phenomlab]

@Panda said in Nodebb design:

One negative is not being so good for SEO as more Server side rendered forums, if web crawlers dont run the JS to read the forum.

From recollection, Google and Bing have the capability to read and process JS, although it’s not in the same manner as a physical person will consume content on a page. It will be seen as plain text, but will be indexed. However, it’s important to note that Yandex and Baidu will not render JS, although seeing as Google has a 90% share of the content available on the web in terms of indexing, this isn’t something you’ll likely lose sleep over.

@Panda said in Nodebb design:

The “write api” is preferred for server-to-server interactions.

This is mostly based around overall security - you won’t typically want a client machine changing database elements or altering data. This is why you have “client-side” which could be DOM manipulation etc, and “server-side” which performs more complex operations as it can communicate directly with the database whereas the client cannot (and if it can, then you have a serious security flaw). Reading from the API is perfectly acceptable on the client-side, but not being able to write.

A paradigm here would be something like SNMP. This protocol exists as a UDP (UDP is very efficient, as it is “fire and forget” and does not wait for a response like TCP does) based service which reads performance data from a remote source, thus enabling an application to parse that data for use in a monitoring application. In all cases, SNMP access should be “RO” (Read Only) and not RW (Read Write). It is completely feasible to assume complete control over a firewall for example by having RW access to SNMP and then exposing it to the entire internet with a weak passphrase.

You wouldn’t do it (at least, I hope you wouldn’t) and the same ethic applies to server-side rendering and the execution of commands.