I see they are using Algolia to do the requests (at least for HN) on the front-end. Unfortunately it is limited, as you cannot get from there whether the current thread is opened for comments or whether the comments are opened for reply. That's why I'm parsing Algolia AND HN's website for each of the comments, to mix the data. Also it makes it easier doing the work on the back-end to add new services and not depending on a 3rd party. See the changelog on the bottom of https://comments.network/ to see those bugs fixed really recently.
I understand, but comments.network js snippet will stop working if you stop your server (comments.network/api), which is a 3rd-party service. Am I right?
Right now (alpha) yes, but I'm exploring other options for its future. Do you have any idea? I want to avoid fragmentation initially so open-source right now is not viable as I cannot support many different pre-releases from different users, but I've definitely considering open-sourcing it in the future ( see http://github.com/franciscop/ ).
For the sake of it I will point out that the alternative, is actually two different third-parties dependencies, http://embedd.io/ snippet and https://hn.algolia.com/ service (;
Since HackerNews itself use algolia, I think its safe to use it. And embedd can easily switch to official HN api.
I just saw that network.comments use internal api too for Reddit while it have a clean json API.
I think it could be nicer for network.comments to be more transparent and to not depend on third party like network.comments internal api. In order to prevent servers maintenance and to give more privacy.
Imagine tomorrow, network.comments seduce every hackers on the world and have more than 100k blog are requesting network.comments/api server. You'll ask guys to install their own instance which is not your first goal to avoid installing stuff.
I agree that Dropbox is nearly same than Google for privacy but it can be a first step to split your data between multiple companies instead of giving all your data to Google.
I know it's totally not perfect, but is it logical?
I'm using Fastmail instead of Gmail. Smartphone app is not really good but you can use any other smtp/imap client.
For android you can synchronize your contacts with "CardDAV-Sync" (carddav is native for iOS).
Fastmail offer contacts and calendars too. Which have a really nice web interface.
I'm still looking for a Drive/Photos alternative. Any advice?
+1 on Fastmail. Been using it for past 6 months with my own personal domains. Actually now like the UI better than Gmail. Such slick, very slide, wow.
Really liking the customizable options for wildcard addresses in particular. You wanna give out an email address for each place you sign up? You can write it as me@site.example.com or me+site@example.com or site@me.example.com - which I find super handy because to my endless annoyance, most sites STILL don't deal with "+" in an email address.
Another +1 for Fastmail. Their IMAP migration tool worked flawlessly shifting multi GBs of emails when I made the switch last month (admittedly not from gmail but from a self-hosted Zarafa install).
Re: Drive/Photos alternative. For the Drive part, if you just need something in the cloud to backup files to (and optionally share them with other people) I highly recommend https://www.spideroak.com. Have been using them for years without issue and they're the only "cloud backup" provider I know of that takes security and privacy seriously.
> I'm still looking for a Drive/Photos alternative. Any advice?
I haven't found any photo storage services that really sell the privacy-conscious angle. That said, I'd trust https://www.smugmug.com/ more than Google, Facebook, or anyone else.
It's like `python3 -m http.server` but in a language that is rarely pre-installed on systems with an external program that has two external dependencies.
Or `http-server` from `npm install -g http-server` in a language that frontend developers and increasing numbers of backend developers actually work with.
That said, HTTP/2 support might be nice to have in Python's http.server or Node's http-server. Albeit you probably shouldn't use http.server nor http-server for production purposes, so it isn't essential.
That's fair. If you could get installation via `setup_requires` to work, so people don't need a (soon to be out-of-date) copy of `requirements.py` in their repo, that'd be a win imo.