/tech/ - Technology★

Very cool. I thought about making a forum for my game server (I have like 200 players, an mmo game...) A chan board and I'll definitely use this.

One of my annoyances with Lynxchan is that when I post or delete the page will load before the change is processed. Is that fixed by your JIT cache?

Also, isn't your JIT cache just a normal cache like they usually work, or is there a distinction I'm missing? I get that it's not like your previous unusual cache.

>>796124

Consider just creating a board on an existing site, like this one (or Endchan if you want Lynxchan).

https://8ch.net/create.php

https://endchan.xyz/

>>796126

>One of my annoyances with Lynxchan is that when I post or delete the page will load before the change is processed.

Yes, that behavior should improve immensely.

I can't guarantee it will be completely gone because the flow of information to cause the cache to be invalidated still isn't bound to the response you get after posting.

But specially on busy sites, like endchan, the main thread will face a much smaller load and you should get an updated page much more often if not always.

>Also, isn't your JIT cache just a normal cache like they usually work, or is there a distinction I'm missing?

Chans usually cache pages once the content change. kusaba and vichan work like that. infinity used to work like that until ron implemented his new caching system after next's dumpster fire.

However, you are correct that more modern general caching systems do work with a JIT cache, like varnish.

>>796121 (OP)

>Advertising all these layering violations as features

>>796139

what is a layering violation?

>JIT caching is caching pages once they are requested, instead of caching when their content is changed.

it is called lazy evaluation you fucking nigger

>>796139

>>796380

Imagine getting mad at irrelevant details like these.

>>796379

https://news.ycombinator.com/item?id=4556125

apparently a design mistake?...

>>796447

imagine irrelevant details being your changelog headlines

>>796701

SAVAGE

ABSOLUTELY BTFO

HUEHUE DOWN

>>796701

>the changelog was about arguing semantics

>the changelog was about muh layering

Autism is one hell of a drug.

>>796722

>being this assblasted

Meguca is a WAAYYY better engine and Lat is a competent coder.

>>796725

>Meguca

If it wasn't a shit livememe site then you'd probably have an arguement.

Blazechan is better.

>>796121 (OP)

>JIT caching is caching pages once they are requested, instead of caching when their content is changed.

Given that most content changes immediately request the changed data, how does this improve performance?

1) User clicks reply button

2) HTTP Post

3) HTTP Get

How does moving the rebuild from step 2 to 3 improve performance when steps two and three occur sequentially for every post?

>>796748

Catalog, previews and board pages.

>>796748

Mind you, this also help immensely when a full site rebuild, since you just drop all caches.

wew lad

What's better for a single-board chan? This, or vichan? I only need webms and images.

Is LynxChan tested on OpenBSD? Because setting up vichan on obsd is a pain in the ass

>>796937

>What's better for a single-board chan? This, or vichan? I only need webms and images.

This. Its easier to install, customize and maintain. Also is configurable to make it having a single board won't feel awkward.

>Is LynxChan tested on OpenBSD?

No.

wasnt it proven that nextchan and blazechan is a lot faster than lynxchan?

kys your'eself

>>796989

1: both are vaporware

2: no, nextchan pretty much died when it had to handle any load and blazechan is built on one of the slowest frameworks out there, django.

When request serving was compared, lynxchan could serve about 10 times what nextchan could.

>>796124

Tell me about your game, why does it have a dedicated chan?

>>796993

>blazechan is slow

proofs?

>>797032

Mind you, I said django is slow. I didn't say blazechan is slow because I didn't test it.

For proof django is slow, this is how it compares to other python frameworks:

http://klen.github.io/py-frameworks-bench/

And here how it east all the ass when compared to frameworks outside the python ecosystem.

https://www.techempower.com/benchmarks/

bump

>>796121 (OP)

>moderation

>bans

FUCKING STATIST SCUM

>>796121 (OP)

>JIT caching is caching pages once they are requested, instead of caching when their content is changed.

So literally a buzzword for lazy builds?

Kill yourself Sergio. Keep your Vichan rewrite to yourself.

Etiquette for talking in a Lynxchan thread

1. Do not call Lynxchan or Blazechan "vaporware"

2. Stop being rude to Stephen or m712

3. "PHP/JS is bad" is not an argument

4. Have good sportsmanship (no KYS or GTFO)

Start a discord if you really want to pick a fight

>>798205

GTFO then KYS

>>798210

GTFO then KYS first.

>>798213

Too late faggot, I already did.

>>796121 (OP)

I don't like JS and won't be using your softwares.

>>798205

>discord

do me a favor and kill yourself

>>798205

JS being trash isn't an argument, it's a fact.

>pls don't be rude to namefags

HAHAHAHAHAHAHAHAHAA

You really should KYS, you don't understand chains enough for this m8.

>>798544

>chains

meant chans

>>798347

What do you think I am trying to imply?

Exactly what you said.

>>798544

Dubs of fail

Bump.

>>807601

Nope, its going through the best beta it had for a while now due to a new site with quite some traffic using it and should release a new version in less than than a month.

I just stopped posting here because I had nothing more to say.

What happened with 666chan?

>>808409

Isn't all that frontend stuff? Clearly they either used defaults or some shitty frontend got popular.

>>807975

No idea.

>>808409

Its the most developed front-end that most people prefer.

>>808427

All kinds of people spam 8ch. It comes with the territory.

>>808434

>All kinds of people spam 8ch

That doesn't make it okay

>>808435

It doesn`t make it my fault either.

https://8ch.net/tech/res/808347.html

Daily reminder not to use (((LynxChan)))

>>808434

>All kinds of people spam 8ch

hex is the spammer tbh

>>808502

You are just looking at the placeholder FE.

Look at lynxtest.operatorchan.org or mewch.net

The front-end is modular and the one I use at lynxhub.com is just meant for development purposes.

>>796121 (OP)

>post answers are not precompiled in html

GNO

I remember seeing one chan that had this, does anybody know it's name?

>>796993

>lynxchan could serve about 10 times what nextchan could

why do you morons always bring this up.

it's almost like you don't know how to benchmark anything.

the whole team behind this website is incompetent, but the fact that you're comparing nodejs serving shit out of memory to PHP templating shit on the fly (with no cache) speaks volumes to lynx dicksuckers competence.

and guess what, when lynxchan was actually benchmarked in a relevant context to infinity next (i.e. templating), the shitty PHP application beat it, and badly, considering that we're talking about request/response cycles over the wire versus a shitty nodejs application templating pages to memory.

I don't have the benchmarks handy because they're lost to time, odili or someone else did them without cripple/josh's captcha fuckup enabled. and even then, the whole point was to cache, the whole caching layer was not done.

you simply cannot compare the two. one rebuilds thread straight into memegoddb for serving (caching), the other did not have any caching at all.

and even then, lynx just seems content to make the same mistakes vichan did (beyond PHP)

>>797040

you have no idea what you are talking about.

>>808615

If by modular you mean custom CSS and maybe a few JavaScript functions, sure, because none of them ever change the templates.

And you still end up with these ridiculous "report forms" you made.

>>808666

any sane schema will store two representations for it, the original markup and the rendered html. I'm sure a lot do that, even next did.

>>808725

>lynx just seems content to make the same mistakes vichan did.

Such was?

>>808726

No, it means that front-end development isn't attached to back-end development. Lately I have worked quite a lot on front-end on penumbra and coincidentally, adding pop up menus like 8ch has is next on my list. I should get that done before the end of the month.

Now, I'm not asking for lenience on the current status of the front-end because I know I couldn't be assed to work on it until now, my point is that because it isn't monolithic like pretty much any other engine, all it takes is someone getting off their ass and doing it and immediately sites can just deploy and use these features.

>>808730

GridFS is not safe for concurrent access, for starters. The problem with vichan was not transferring furry porn between servers, Lynx.The problem was NFS shitting the bed at first, wherein they had to switch over to FreeBSD and get two FreeBSD experts (in-house and weev, iirc) to help tune the shit, and it continued staggering at this pitiful load. Then came the statistics problem, a PR that both codemonkey and hotwheels greenlit and tried to blame entirely on josh after they just had their little shitshow, which only added to the problem.

The problem was locking and hanging. And PHP. You specifically chose an abstraction (over dogshit, I might add) that is not safe for consumption in this manner and put a dumbfuck slow as queue in front of it which I imagine, beyond your dogshit JS templating library, is the reason that lynxchan simply cannot template pages as fast as an over-engineered PHP webapp.

And no, the worst case scenario is not "page shows up half templated." It is (was?) broken, like vichan was broken.

>>808739

>GridFS

>NFS

kek. just use amazon cloudfront

>>808741

epin

>>808739

Also, I don't have time to actually read undocumented Hasklel code (like pretty much everything in that ecosystem) but I imagine codemonkey just implemented a shitty version of redis and called it a day. That's really all it took at this load.

And to end, if Josh had not fucked up and been so gung-ho on his Plan A, along with letting a fucking captcha be generated on every new session OFF THE REQUEST CYCLE, you wouldn't have had the opportunity to be a humungous faggot in IRC and spam DoS hammer benchmarks at a PHP server with no caching and trick retards into thinking that you know what you're talking about. Web dev is dogshit, but it's not hard to understand these basic fucking concepts.

The fact that you continue not know how to even benchmark after how many years shows you're either disingenuous, or a fucking retard.

>>808743

>epin

not an argument?????

>>808739

>wherein they had to switch over to FreeBSD

I'm done.

>>808746

cloudfront is a CDN, moron, like cloudflare where invalidation requires 10 minutes at a minimum, and you specify the caching behavior which doesn't really help when we're talking about content that's always changing because the caching behavior would be little to none on boards, no better than cloudflare in the context of this website. and as far as I'm aware no ESI. if you wanted to shitpost, you should've said akami.

>>808758

>moron

nice ad hominem. not an argument though

>when we're talking about content that's always changing

you were talking about furry porn though

>>808765

static content was not the problem, that's what lynz thought the problem was so he put it all in one big "schemaless" (more like dynamically types schema) pot, posts, images, media, and settings and let a filesystem abstraction over memegoddb deal with the problem because he's lazy.

>>>/v/13657495

StephenLynx is attempting to scrape 8chan, put the result into his shit software and monetize the result. Thoughts?

>>810140

cool

>>810140

I think that's completely fine as long as you only scrape boards that are in https://8ch.net/8archive.json.

>>810140

>StephenLynx is

The mewch staff is.

I have nothing to do with that.

>>810140

>mfw i managed to do it with infinity next before lynx could

:^)

>>810267

I've scraped content from Lynxchan to put it back on Lynxchan back in July, for what it's worth

>>808739

>weev

>expert

stynx

>>810267

So you are Josh?

>>817534

So you are retarded?

>>818358

>no rewrite in rust announced yet

I'm starting to get slightly worried. When are going to RIIR?