Just before you, my fellow system programmer, scroll past this, let me say this:

🍬 The web is actiually simple. 🍬

Both HTML and CSS is declarative. It's all easy when you understand the concepts, learn how to be idiomatic and quit trying to do that imperative bullshit in languages that aren't imperative.

HTML is simple. You know the boilerplate: doctype, head, body, that's all. Just mark it up and do NOT look at it before you end, mark it up as it were article or something. The appearance is up to css.

CSS is simple. You may even forget bem or rscss, you're already a skilled software developer. Use common sense and your code-splitting and naming skills you gained reading The Code Complete or doing software development for years.

Forget mockups. Forget absolute positioning, forget setting width and height in pixels. Go to awwwards, find some inspiration. Draw some buttons and fields on paper with your good old pencil. Then go and write some css. Feel free to steal some shadows and transitions from codepen.

Read about 8-pixel grid system. Let every element push away from others by setting something like margin: 16px; and whoops! You've just got fully responsive and got great vertical rhythm without even using media queries!

Oh my god, do NEVER set width and height explicitly! Type something like button { width: 120px; } and bang! The entire web page is broken. Quit that shit. Let it resize as it should. It will resize itself to fit its contents.

HTML is by default ready for your template engine. That's how you receive data from server — as server-side rendered, plain old HTML page. On the other hand, the form element is the most axiomatic and simple way to send the data to server. That's how you send it — as plain old GET or POST that every webserver can handle.

All of there are true:
1. It's easy to get great 100% responsiveness without media queries.
2. It's easy to align items in row, it's just one line of css. Maybe two, if you still want elements to wrap, but want to use flexbox:
.parent {
display: flex;
flex-wrap: wrap;
}
3. HTML and CSS are fast by default.
4. You don't need mockups to achieve great visual experience. Mockups is imperative, web is declarative.
5. You may not even need JavaScript to make great website.

Go on, ask me a question about web! I'll ready to answer everything.

It all started with an undelivereable e-mail.

New manager (soon-to-be boss) walks into admin guy's office and complains about an e-mail he sent to a customer being rejected by the recipient's mail server. I can hear parts of the conversation from my office across the floor.

Recipient uses the spamcop.net blacklist and our mail was rejected since it came from an IP address known to be sending mails to their spamtrap.

Admin guy wants to verify the claim by trying to find out our static public IPv4 address, to compare it to the blacklisted one from the notification.
For half an hour boss and him are trying to find the correct login credentials for the telco's customer-self-care web interface.
Eventually they call telco's support to get new credentials, it turned out during the VoIP migration about six months ago we got new credentials that were apparently not noted anywhere.

Eventually admin guy can log in, and wonders why he can't see any static IP address listed there, calls support again. Turns out we were not even using a static IP address anymore since the VoIP change. Now it's not like we would be hosting any services that need to be publicly accessible, nor would all users send their e-mail via a local server (at least my machine is already configured to talk directly to the telco's smtp, but this was supposedly different in the good ol' days, so I'm not sure whether it still applies to some users).

In any case, the e-mail issue seems completely forgotten by now: Admin guy wants his static ip address back, negotiates with telco support.
The change will require new PPPoE credentials for the VDSL line, he apparently received them over the phone(?) and should update them in the CPE after they had disabled the login for the dynamic address. Obviously something went wrong, admin guy meanwhile having to use his private phone to call support, claims the credentials would be reverted immediately when he changed them in the CPE Web UI.

Now I'm not exactly sure why, there's two scenarios I could imagine:
- Maybe telco would use TR-069/CWMP to remotely provision the credentials which are not updated in their system, thus overwriting CPE to the old ones and don't allow for manual changes, or
- Maybe just a browser issue. The CPE's login page is not even rendered correctly in my browser, but then again I'm the only one at the company using Firefox Private Mode with Ghostery, so it can't be reproduced on another machine. At least viewing the login/status page works with IE11 though, no idea how badly-written the config stuff itself might be.

Many hours pass, I enjoy not being annoyed by incoming phone calls for the rest of the day. Boss is slightly less happy, no internet and no incoming calls.

Next morning, windows would ask me to classify this new network as public/work/private - apparently someone tried factory-resetting the CPE. Or did they even get a replacement!? Still no internet though.
Hours later, everything finally back to normal, no idea what exactly happened - but we have our old static IPv4 address back, still wondering what we need it for.

Oh, and the blacklisted IP address was just the telco's mail server, of course. They end up on the spamcop list every once in a while.

tl;dr: if you're running a business in Germany that needs e-mail, just don't send it via the big magenta monopoly - you would end up sharing the same mail servers with tons of small businesses that might not employ the most qualified people for securing their stuff, so they will naturally be pwned and abused for spam every once in a while, having your mailservers blacklisted.

I'm waiting for the day when the next e-mail will be blocked and manager / boss eventually wonder how the 24-hours-outage did not even fix aynything in the end...

Ugh. That may have been a mistake.

I'm deep in a large effort to refactor my project. It's a one man deal and something I've been working on pretty much every day in some fashion for nearly 10 years (five years ago I started a scratch rewrite to move from a fully CGI server rendered application to a browser rendered asynchronous version built around JS) and that took me three years.

I started this refactor about 8 weeks ago. Turns out I've been tackling the largest modules and progress has been decent. So that's good.

But I got to wondering ... Just how much code is there?

So I whipped up a quick script to do some calculations. Read each file and get a line and word count, skipping empty lines.

In JS it turns out I have 83,973 lines and 467,683 words.

On the back end, 86,230 lines and 580,422 words.

Average publishing stats say the are about 250 words/printed page.

That means I'm confronting refactoring 1,870 pages of JS. That's the size of several decent sized novels. (I think I've done the equivalent of Maybe 400 at this point).

Makes me feel like the walls are creeping in to know how much is left to go ...

the more i learn about web dev, the more i realise the reason for its mess up . There are 2 major problems in it : the people who create various important concepts and tools for web dev were 1) working on it without any collaboration and agreements on the philosophy and 2) were too stubborn on their ideology i guess.

There is no limitation to anything's functionalities, and the limits that are "defined" are badshit crazy. for eg:

====================================
HTML creator : "I am gonna make a language that would provide a skeleton to web page. it will just have the text and basic markers to let the scripting and styling engines/languages know which text is supposed to be rendered and how.
It won't provide any click or loading functionality.

someone: "So i guess opening a page or loading an image would be handled by JS or other programming language? also, bold , italic or division would be added via CSS?"

HTMLguy : Nah, my html engine would ALSO do that.

someone : what , why? won't that just be stupid and against your philosophy?
HTMLguy : WHAT? am too awesome, can't hear you
w3c , 50 yrs later : sorry can't change this, gotta support the 50 yrs of web dev and billion sites

=================================

CSS guy: I am gonna make the world's best beautifying stylesheet language to provide colors, styling, fonts and backgrounds to a page. every loadings and clicks would be handled somewhere else

Some1: cool, then clicks, hover and running of animation would be handled by JS only

CSSguy :Umm, i guess i could handle those.

Some1 wha-?

CSSguy : Thankyou Thankyou Thankyou for the nobel price!

====================================

JS guy : I am gonna make a god web programming language! It can do everything: add/remove html tags, add styling, control animations, control browser, handle clicks , perform operations, everything!

some1: cool! you must be making very large programming language with lots of modules.

JS guy: No! i am gonna keep it small. no built in classes and file imports! just use the functions directly. if someone wants the additional lib functionality, install them on your server

some1 : innovative! what's typeof NaN ?

JSguy :shut up.

Why the fuck nobody talks about Multi-page apps?! We went from a Web where everything was Multi-page server-rendered, and now everything for Web developers is "Single-page apps".

What about websites who can't do that? Not everything can be a single-page app. Only my uncle's restaurant website, or something which is TRULY a full app. No half choices.
If your website is a multi-page app/portal which actually PRELOADS data, instead of doing 100 fetch to an API within a page that is full of loading bars, well, your life is a pain.
When you want a first contentful paint which isn't a white page, well, your life is a pain.
What are React, Vue, Ember, Angular (let's exclude Svelte and Marko) going to do about Multi-page apps and SSR?

React-router sucks to me. It's performance is weak and it's useful only when you have an SPA with multiple sections which can be treated as pages (e.g. A single SPA divided in tabs).
Server-side rendering is the worst pain ever made by humanity, in React (and prob Vue, I didn't try but I can bet). And even when made easier from libs like Svelte and Marko, I (personally) can't get it to be faster enough compared to a traditional website without a JS framework and with a templating engine.
Anyways, if there's anything that I learnt from React, is to stay away from Next.js. Perfect, beautiful, mess.

All JS frameworks just seem to bloat the code and make it worse and slower, even though they're REALLY helpful.

Why? Why everyone loves them if their downsides are so clear? Why 3 projects out of 3 I made (1 React SSR, 1 Vue, 1 Marko SSR) are and will stay painfully slow and bloated, full of shit, even if in 2020 we should have evolved with the famous three shaking, with the famous lazy loading, etc.?

I am just frustrated.
And let's not even talk about Webpack, Rollup, Lasso, those module bundlers shit which are harder to configure and understand than finding a needle in a haystack.
Lasso was the easiest to configure but I anyways can't understand it. Webpack seems it was made to handle SPAs, as any tool in this freaking world, and not even considering an easy way to integrate multiple bundles for multiple pages (I know it's pretty easy, but with component sharing between pages and big unique bundles Next.js handles it soooo bad it feels like hell).

Am I the only one?

Sorry for the long rant. I just needed to rant right now.

Just had an argument (with myself ;)) whether to write a completely server rendered site or a js app.

The only stas I have manged to find is from 2010 saying potentially 2% of users could have js disabled.

Why cannkt I find newer stats :/

jinja templates make me look towards html in a whole new light. are we 'inserting' data to an already rendered page? am i really mixing server code with ui ? It doesn't feel so. there are if else and loops being executed for html code, like wtf?

I don't know but everything feels so good. like i was literally hating every piece of website i was writing in php. everytime i wrote <div>....</div> followed by <?php ... ?> followed by another html tag /php tag in a fuckin php file, i wanted to kill someone from w3c.
WHY THE FUCK ARE WE ALLOWING THE MIXUP ?WHY IS PHP FILE HOLDING HTM TAGS? WHY?WHY?WHY?

But this... this is beauty. their is separation of concerns. jinja has some big powers, we can loop, repeat, make clauses, inherit other html classes, load html content into blocks, set variables,
but main concepts like file handling, response/request handling,calculations,etc are all being done in separate python files. I know that these jinja templates also might be running python in background, but atleast a developer cannot fuck up that code.
we can be sure that if correct jinja codes are written in html, then it would load correctly. And wherever devs doesn't fuck up, the output is better to understand and more maintainable/scaleable