the download link there is broken seamonkey-project.org/dev/nightly

Sompi: the correct link is in the topic

reporting the issue to people should really be archive.seamonkey-project.org/nightly/comm-253-latest

WG9s: how large is the entire at least RELEASE versions of seamonkey?

would be a question for ewong: all i know id what is in archive.seamonkey-project.org/nightly/comm-253-latest

we should really rig up your build system to produce xz compressed tarballs

so the 1,52 nightly builds i do I con;t do all the languages I only do de and the rest langpacks.

do de jsut to aid frg

well that was one solution I gave to the issue of loosing mozinfra

just a couple of guys doing builds

lol

ewong does the official release builds

yeah but you don;t have a tinderbox fleet anymore

like in the old days

solution, couple of guys doing builds

not anywhere so i just check what happend after the build completes or does not

nothing that goes red when a build fails other than I get a tst on my cell phone

no tinderbox builds or tinderstatus add-on that puts a red box on my statusbar

kist i do builds and if they fail i try to fix

othersise up to whoever posted the last patches has to fix et

etc

works only becoause just three of use posting anything

there are 3 others contributing but frg seems to be doing the posting of their stuff

this is on 2.53 no one is really posting anyting on central expect me only fixing build bustage

Well then certainly time for a change.

i have put my thundebird stuff on hold until I build a basic custom element cause there is some piece of it I am missing that I know would tie up the concept for me.. just gonna have to go through creating one step by step

gonna do that maybe tomorrow

today I am doing server stuff and my xre app stuff which in turn illuminates central for me and as I explained earlier if you wanna chance to run the suite on central you need a piece of it that works with devtools and that is exactly what I am doing for my example app. So I figure add a "reconstruction" component based on it that can also spawn all the various windows for fixing up.

if you are serious about trying to get central actually working bug 1452448 at one time had most of the things needed but i don;t think it is still being maintained. i would do a linux build with debug enabled and then fire it up from a terminal window and look in the window for etrors encounted during startup as good places to start.

I am serious about everything I wanna do.. BUT I hold no expectations of making it viable but if it suddenly became as such so much the better.

It would be stupid to get hopes up on it.. BUT better central functioning will allow better compairitive testing and faciliatet backports

Right now I am still experimenting

well other that what i just said I am not in charge here. you need to be yalkting to either frg or IanN

I mean I made a main window spawn with an e10s browser that goes to a site

lol

Sompi: i asked to have the link changed

WG9s: yeah my main aim on suite on central is maybe have that as an option in the list of choices for you guys if possible

i am not gonna try and railroad you guys into doing shit

that has NEVER worked

lol

But without an overpaid team of crack modern developers I think the add a piece in that works and spread out from there is the best and simplest option almost totally originating out of stuff I am doing anyway

Cause MINIMUM suiteglue, browser content handler, and devtools init on navigator.xul has to be done before you can do anything otherwise

but all i do is builds. only people who can r+ a patch are frg and IanN

why I said you need to talk to them and not me

well yeah I have to one.. do it, two organize it into consumable and nicely patchable bits, create bugs or "throw it all on frg's desk" and hope for the best lol

only thing i can add to the gitlab aueue are build bustage fixes

I wish I could help on the webcompat bugs

but even if I did it would be outside the patchqueue

if we could get the UI to actually appear that would probably be changed to I can also fix startup issues

WG9s: one preference can make navigator.xhtml show UP but it won't do nothin but be a few toolbar border lines

the reason it doesn't is because suiteglue is an xpcom js component NOT a staticly registered xpcom jsm

thing

so it can't run

so it doesn't do dick let alone spawn the main window

using the toolkit.defaultChromeURL preference you can get the toolkit default commandline handler to spawn any winder you want

but that won't do much without devtools

now knowing that and fixing that are two way the hell different skill levels

lol

WG9s: i am not gonna lie.. this is like making the jump from 1.8.1 to gecko/24

the TRADITIONAL way would be to follow each and every mozilla release up the line fixing everything and then just slapping that over central. But that may actually take longer than fixing central in place might

at this point

you know Thunderbird DID convert the customizabletoolbar binding to a customelement tho still xul based so there is that too and they have a statusbar

I dunno it is exciting to me to wanna give it a shot even if it just improves backporting

I want to compile Seamonkey to non-SSE2 targets

Currently rustc does not allow that

Some time ago I saw some dude that made a Rust compiler backend for Win9x, which also means "pre-SSE2 targeta"

can't recall it now, but I guess a bunch of patches and stuff are involved, so nothing you're going to find on upstream Rust

Sompi: have you ried uilding sith any of the i586 targets? i586-pc-windows-msvc i586-unknown-linux-gnu

I haven't got the build process working at all

I tried to do it based on the Mozilla documentation but the build scripts had some hardcoded paths and it didn't work on Slackware

Sompi: why focus on that when webcompat is an issue

I also tried to compile Firefox

what is the benefit to them to support old computers when webcompat is a lingering issue

I mean non-sse2

that's 15+ years old i bet

you can't expect today's software to run or run well on ancient hardware no matter if the UI happens to fit in with the era

What is "webcompat"?

web compatibility, able to go to websites and have them work in the browser/navigator

I occasionally used a Pentium III computer to browse web and it worked just fine with all sites that I use, but then Seamonkey and Firefox started giving "incorrect instruction" errors because rust adds those SSE2 instructions there

well the web as a whole has made pentium III a no go

this isn't the world wide web anymore dude

I wrote a javascript game that works just fine on that computer

600 MHz Pentium III

this is the openweb of closed proritary webapps

Or worked, when Seamonkey still supported that CPU

ContributorTobin: If I have a i686 package of some program, I expect it to work on Pentium Pro

that's not reasonable

dude

Rust devs are assholes and they break their compiler on purpose

It is reasonable to assume that i686 means Pentium Pro

It has always meant it

Why have 32-bit builds in the first place if you only support Pentium 4 which is already 64-bit

pentium 4 is not 64bit

Prescott is, and the other versions are slower than Pentium III

ok

and no one can realistically support hardware that old

It is not old

it's 2024 dude

And it does not need to be "supported"

The idea of public source code is that the program can be compiled to any target

.. kinda does by every lib compiler and dep that goes into compiling seamonkey

Seamonkey is not written in assembly

some of it is

What modules?

there IS assembly in the tree

Sompi: go find it.

there is assembly, c, cpp, xpidl, ipldl, webidl, javascript from es4ish to latest there is xul, html, css, python, perl, bash, maybe even a lingering batch script..

It is a web browser and supposed to be multi-platform

Sompi: what os is you issue building for non-sse2?

it's an internet application suite

actually

WG9s: I tried it in Slackware

hmm well not sure if they support rustup and adding targets

What linux distros should work?

Sompi: Multi-platform hasn't meant multi-architecture outside arm and recently with risc-v for like 15 years cause no one cares about old processors nor failed lines such as itanium

ContributorTobin: If the program is written in high-level languages, it should compile just fine to any target

multi-platform means windows, linux, macos

The rust people are just breaking target compatibility on purpose and it will destroy Linux

Sompi: you're clearly not listening nor willing to learn anything about what you are making broad, outdated, or incorrect assumptions about dude. It's not good for you.

but you might try adding ac_add_options --target=C to your mozconfig this might also require doing a rustup target add i586-unknown-linux-gnu first

I don't have money to buy new computers and also I don't want to produce electronic waste

WG9s: Thanks, but what linux distros are guaranteed to work with the toolchain?

No, WG9s Sompi knows exactly what he is talking about and surely does not need any advice. They are good to go and know more than we do about this codebase. It's cool. Off he goes!

I need to set up a virtual machine for compiling that it

Sompi: traditionally rhel or centos always works. just might not be the latest if it just came out.

ContributorTobin: well then perhaps I will just block both of you

WG9s: please don't

sorry

Does the current rustc even support 32-bit targets properly anymore?

32bit is gonna be harder for everyone to keep going just alone

I read that they completely dropped the support for different 32-bit x86 targets and now the compiler just can only produce 32-bit code for new x86 CPUs when asked

The switches for i686 and older were broken and Debian maintainers shipped their own patched version of rustc that used the same naming conventions for targets as gcc does

Sompi: I am sorry for jumping at you I didn't intend to do so.

And I tried to compile that patched version of rustc myself on Slackware, but the after compiling the toolchain detected a mismatch in some checksum and just downloaded the stock version of the compiler and replaced my modified version with it. :(

ContributorTobin: It's ok

Not really but thanks

(being ok)

I'll try compiling Seamonkey again on Redhat some day

on some el flavor it is pretty easy

but i dunno redhat has dropped 32bit as well

everyone is

If I knew how to cross-compile I'd tell you to try that

I been trying to work that out for years

apthat is what I tried to tell hum when you shut me off

Of course it would be cross-compiling, so the architecture of the host machine does not matter

I am sorry again

ContributorTobin: If you look up the most popular linux distributions from DistroWatch, most of them still support 32-bit x86

unfortunately latest rust does not support building for anythong older than i586 which migh not have been sufficient

32-bit x86 is not even rare. Many much rarer architectures are also supported

WG9s: What do they currently mean with "i586"?

i have no clue exactly how that differs form i686

was really just a shot in the dark

i586 is likely orginal pentium with mmx

For gcc and practically every sane compiler i586 means Pentium and i686 means Pentium Pro, Pentium II and generally any 32-bit x86 CPU without SSE2

i would think either i486 or u686 is shat might be needed herei think for this i586 means pentium and i686 mans current intel

but 32-bit

mode

i686 was the last ix86 ever used

so x86 but 32bit is how I have seen it used for well 15 years

ContributorTobin: AFAIK gcc does not produce MMX instructions by default when i586 is used. You need to use an extra flag to have MMX instructions

but for targets in llvm and therfore rust i686 is kind of using a new 64-bit ptocessor running a 32-bit app

so I thought i586 target might fix his issue

The thing is it is getting to the point where features needed do not exist on older processors and the cost of maintaining that code is large especially when everyone seems to want off intel x86 based anything in the industry or so they make it seem. I am betting on RISC V my self becasue a future that is all ARM isn't a future I want to live in.

I am not sure even UXP can be built with an i586 target successfully

IBM PC compatible machines with traditional BIOS were/are well defined and documented. Easy to write even operating systems for them so that it works on every computer

and it is plain gcc

Sompi: uefi is well spec'd and documented as well but no one follows it properly

ARM on the other hand is a complete opposite of that. And also modern UEFI-based PCs, basically the OS developer can assume nothing. The ACPI tables are complete crap and they are so on purpose, because Microsoft wanted it that way

oh well i gave up long ago on building seamonkey using gcc

WG9s: uses clang now everywhere doesn't it?

yes

well winders is a bit fuzzy but

cause all them vs components

i did switch to gcc fo my 64-bit linux builds on 2.53 becaue the mach buildsymbols things works better. but use clang for my 32-bit builds becuase they fail using gcc

sm is clang, mozilla is clang, I guess by extension I am clang too.. everyone is clang. *hits metal pot with a spoon* CLANGGGGGGGGG

We need a RISC-V system with a BIOS-like well-defined interface in its ROM

Sompi: design one ;)

so my 32-bit builds work but if they crash less sybols available to the crashreporter site

UEFI may be well documented and specified, but it is very badly designed

did you know the uefi standard requires that regardless of CSM the first sector of the efi partion must be allowed to direct boot

but no one supports it

hmm perhaps i dhould just build me 32-bit builds with strip disabled so that is not an issue.

ContributorTobin: If the "compatibility support module" is used, it should just load the contents of the first sector to address 0x7C00 and jump to that address, no?

or disable crash reporter

And not care about what's in the sector. Only check that the boot sector signature 0x55AA is there.

well that is even lamer so you have to debug locally so I sould have to disable sytip and enable debug

that isn't what I am talking about.. efi allows 1st sector native booting from an efi partition

without bios emulation

and that would mke the app bigger so less likely to work on an old 32-bit system

I know that many modern BIOSes, for example Insyde BIOSes, don't work that way. They refuse to boot if they don't see a MS-DOS/Windows compatible partition table there

but no one impls it or if they do it is busted and no one does it

ContributorTobin: Ok. I did not know that. Most UEFI implementations refuse to boot at all if there is no GPT partition on the disk

one good thing about building using clang is that as long as you are stuck using rust both are just front ends to where the real work is cdone and both use llvm so beter cmpatability between the c and rust code

I thought that they need the GPT partition table and a FAT32 partition there, where the EFI binaries are

rust clang and glan++ are all just front end parsers the real comiling for all three is done in llvm

Sompi: btw, Rust was created by mozilla and has a mozilla atmosphere even today do you HONESTLY expect Mozilla to keep supporting old cpus forever or push eeveryone to the latest by any means possible?

or not push*

WG9s: So is that why Mozilla switched, better rust interop?

i am not sure

I still feel it was a dick move to completely reformat an ESR tree mid-stream with clang formatting

basically makes diffing impractical

Mozilla is moving farther and farther away from free software ideals

ws not involved in that but seems on real compiler with just the front end parser front end being different sounded better to me for interoperability

Mozilla is a more complex situation than I may have portrayed in 2017 dude

and they have a lot of problems, most of which seems to be reactionary

A relatively modern browser for DOS would be cool

like how Firefox the UX has stagnated and the solution is shove in AI

seems like linux is more interested in rust than before instead of just aing thy would support drivers written in rust, now seem leaning toward rewriting parts of the kernel in rust

WG9s: we all saw that coming didn't we?

WG9s: Probably not going to happen before gcc can compile rust

so will anyone hardfork the kernel make it rust free?

lol no

The current official rustc compiler is in hands of people that purposely want to break stuff and rewriting static parts of Linux with rust would just destroy its portability

i hAVE NO IDEA JUST READING NEWS ARTICLES

oops sorry hit caps lock

THERE IS NOTHING WRONG WITH THE VOLUME OF MY VOICE!!!!!

lol

Many "news" articles nowadays are just hyping nonsense

Sompi: nowadays?

But gcc is going to have rust support soon

let's just rewrite seamonkey in objective-c and be done with it

And it will fix the issue with the targets

no it isn't

it's cpp for javascript coders

who were not trained properly

strangly i don't know it

anyway biggest issue i have with rust is not with rust but is with mozilla. mozilla has these crates they provide to do things and they depend on mozilla added temp exceptions to the rust rules to make it easier to transition a mozilla module to rust and then they fix it all to actually comply with rust one transitiond and then the temp crap they did is no longer allowed in the next version...

...of rust

was only evfver allowed with warnings that for a non-mozilla build would be errors

I almost started learning rust, but then I noticed that most rust advocators are shitheads, and as I wrote in the previous sentence, advocators. For them it is an ideological thing and they have long time ago stopped caring about the fact that things should also actually work and not just be ideologically pure

but that is why we have issues with seamonkey becuase we backport mozilla bugs that reauire the kluges and then update to a new version of rustc and suddenly it breaks

A memory-safe low-level programming language would be a very good thing to have. Too bad the state of rust is what it is.

that is not rust's fault that is mozilla's fault

WG9s: yeah, do you not have comm's rust workspace crate hack shit

I didn't check when it was done on central

anything in comm-central i have

but not in comm-release used for 3.63

2.53

cause and I am no expert, yet, but it looks like they use a hack to basically supply their own crates and include mozilla ones and generate an alt geckorust thing

Sompi: likely won't change until it is widely adopted and enough people at the same time come to the realization that it is incredibly broken and then they fix it which busts everythng before it and a whole new group is pissed off about it

rust is perfectly named.. the question is will the corrosion eat through everything before it is repaired

but 2.53 builds with latest rust. we don't use it for macOS because rust versions past 1.73.0 don't support as much backwards comparability as we ate trying to provide

yeah that is all new to me

or compatibility even

I know it exists and I can point to it

I can write programs in C. I can also weld metals.

that's about it on the rusty front

so far anyway

I can fix damages that rust has done.

Sompi: I wish I could write shit in C/CPP

I understand C++ but I don't memorize enough of its things that I could actually write it. But everything C should also be valid C++ so technically I can also write C++. :D

i wish i could go to CDC3300 assembly language

oh that is kind of how I do c++

I did a little 68k assembly when I was hacking on a Sonic 2 for the megadrive disassembly

does that count?

Every instruction set has its own assembly

yeah i know

x86 is incomprehensable except for what looks like 68k

lol

c++ is mostly c with the idea of new to crate something that stays as long as refernced and addref to add additional refernces as far as I know

and memory shit

in str8 c you have to manage memory yourself

68k assembly has a cool way of telling the addressing modes. On x86 you have at least three different syntaxes that are common and on all of them the addressing modes are often very vague

that is what i mean c++ is so pwople who have no clue how to manage memroy can write programs that son;t leak like a sieve

.net teaches people to expect it and tout it as a feature

or did for most of its run

lol

but encourages people who really don't know how to program to attempt to do so

every .net thing I wrote fuckin leaked even simple bs like a tray thing that pings a site every five minutes and loops a wav file if it can't

just sit there idling eating bytes of ram

after a few hours it eat some megs after a few days it at several hundred megs

I'm not sure but it seems that the rust's equivalent of "crashing"/segfaulting is that it starts eating up memory like crazy and finally the OOM handler kills the program

I very much prefer segfaults over that

well we all fucked around and didn't help stop rust when we had the chance, we are stuck with it.

might as well learn to manage and control it so it doesn't manage and control us

ContributorTobin: btw you asked about the xref wanted to thank you for keeping the ghostbusters reference alive

ContributorTobin: Diversity is generally good in computing, and also in programming languages. Too bad that rustaceans actually seem to be against diversity, they want everyone to use only their thing

Diversity is good for freedom

Sompi: you must be a democrat

I don't live in US

depends on his definition of diversity doesn't it WG9s

I meant technological diversity, not ethnical diversity. Ethnical diversity is just a neutral thing, not a good and not a bad thing either.

are you filling a need or a quota, that is what matters to me

I know but republicans seem to think Either you are a WASP (White Anglo-Saxon Protestant) or you should go back to where you came from

you might not be familiar with the WASP term was somthing form back in the 1960's that we thought we had ridden ourselves of

now they call themselvs christian fundamentalists

well the world sucks too much for there to be a christan god.. that old testiment fucker tho.. not so sure about him.

Corrupted demiurge!

the new testamant at least had this idea that you should lobe your enemies. strange as it ws then and now

love not lobe

yeah and the ones who are loudest about inclusion and diversity will find every single thing about you to reject you out of hand

same shit

humanity keeps making the same damned mistakes

well you also get the NIMBY thing i love those others but Not In My Back Yard

I remember when George W. Bush talked some crazy stuff about the axle of evil and wanted to bomb the whole middle east, and back then democrats were anti-war. How things have changed...

but was a different war now i don;t see why wee are supporting Isreal.

i dunno but it sure feels we skewed into some tanget that keeps getting worse and we keep surviving it

like this today is not the world anyone was calling for

anyone outside a power structure with wide sweeping reach anyway

Gworce W BUsh wanted to go in and finsih what he thought his father left undone. Issue is, go figure, Dadham Hussein was actually the one keeping stbiity in the regieon. There never were any weapons of mass destruction, but he insisted he had them and that is what kept his neighbors form making war.

Well W also taught us the internet is a series of tubes

Now democrats want to bomb half of Ukraine to drive Russia away from there

I stopped paying attention.. We may get nuked at any time and there is a chance it will be from our own government.. so.. that doesn't consern me because unless we get nuked I still have to deal with things as they are atm

also the idea he was hacking off body parts of citizens. those stories is what kept the factions form in Iraq from fighting against each others. He ruled with an iron had and feared by his citizens and neighboring countries, but it worked.

not saying he was a great guy or that i thought his methods were the greatest. but they seemed to work to keep peace until we came along

War propaganda is easier to identify afterwards

we could have launched a tactical nuke at his palace during some speech and it would have been less casualties over all

Today's news are tomorrow's war propaganda...

but oddly he kept peace in the region and once we got rid of him look what happened.

WG9s: Same happened with Gaddafi

And all those refugees come to EU & Russia, destabilizing and weakening them

has it occured to you that while the democrats WANT war they SUCK at war but if they START MORE WAR and loose then they instantly can turn back around against the republicans because we DO know the republicals love war it makes them money with military industral complexes and the like

and if that happens

And I think it is also part of USA's plan

the democrats can instantly switch back to anti-war

and no one will ever call it out

it's actually quite amazing objectively speaking

it is the republicans wo are pro was they thing putin should be able to take over territory using his miltary force. you can;t get much more pro war than that

frg_Away looks at the backlog and then decides its just not worth it lol

I am out of the loop till at least tomorrow and just ocassionally reading stuff.

we are managing a non-heated politcal discussion that somehow happened.

lol

I DIDN'T DO IT lol

ContributorTobin: you wer the one to blame democtrats

Anyone wanting non sse2 or old i586 support today should see a shink :) 99% of todays websites will crawl on these old systems. Core2Duo is borderline only. WASTE OF TIME AND RESOURCES. Pick and old version and use it if you must.

ContributorTobin: is being ignored by me

frg_Away: Actually Pentium III has about the same count of instructions per cycles as Core 2 has

I don't know how it compares to those modern ARM single-board computers. At least the first and second version of Raspberry Pi were a lot slower than Pentium III or even Pentium II.

The ARM world is weird. Modern smartphones should have a lot of computing power, but they are slow af in everything

off topic on

It is not my country but the republican party seems to act like a cult these days and anyone will suffer if Trump wins again. The Dems are not my cup of tea either and Biden is too old for another term. But not my country. Only one thing is iportant Putin must be stopped or we will see more misery down the line.

off topic off

Well I tried to tell you how to see if building for an i585 target would fix your issue but Tobin sjut me down so if you prefr to listen to him I am giving up also

frg_Away: if biden is too old so is trump so why do we have a choice betseen too men who are too old?

Modern Android does not even have proper multitasking. Only the app that has the visible "window" gets CPU time, everything else is paused. It multitasks like DOSSHELL from MS-DOS 5.0.

And still everything is slooooow.

Sompi. You need clock x64 4-8GB and 2 cores at a minimum No P3 is capabale of running a modern browser in anything but self torture mode.

frg_Away: I used it and it worked until rustc broke it

I even wrote a javascript game that draws things on a HTML5 canvas, and my 600 MHz Pentium III ran it just fine and the framerate was fine, it was playable

and still I think it is self torture. I amy using a used $60 T430s from about 2024 right now and it is great. SSD not included in the price but the rest is.

about 2014 not 2024

Pentium III has basically no difference to new CPUs with well-written programs

Sadly most JavaScript code is not so well-written

Sompi yes The P3 is not bad but it is completely obsolete. Only good if want to put together a system playing old games or very special hardware wih no later drivers.

I still use a T42 P3 for programming eproms.

The computer in question has a parallel port and I need it for printing

And sometimes I need to use the web browser but Seamonkey does not work on it anymore

I use a Color Laserjet 4550 from 2000 and was able to put a network card in. There are later i7 or Xeon 5600 boards with a parallel port available too and they are ok too.

And the computer also used to be able to do certain things so it should still be able to do them...

Sompi yeah probably nothing current will work on it. But that is as it is. To make it work you woulld need too much time and resources.

Don't listen to me.. no

damn it

Sompi: please listen to WG9s, I was out of line when I said that shit I said cutting him off.

fuckin politics.. just ruins everything. Have a nice day everyone sorry for the disruption.

I need to try cross-compiling Seamonkey on a Redhat Linux. Right now VirtualBox is not working on Slackware because the working kernel driver is not fully open source. If they don't fix it soon, I can always use QEMU or something

Or just install Redhat on some physical computer, but I don't have many 64-bit machines

Note to self: Say less things and make what is said more helpful...

It is also a problem if compiling the program requires a very specific type of computer :(

Sompi: i tried to help. you preferred to listen to Tobin. I susgest you ask him

Sompi: If you are gonna listen to me or whatever, I recommend a little less output and little more input from WG9s specifically. I am gonna do the same.

WG9s: I will try what you suggested

I mean what you suggested earlier

A friend also needed a Seamonkey or Firefox build without SSE2

0225|18:05:10 < Sompi> UEFI may be well documented and specified, but it is very badly designed <-- years dealing with BIOS and El Torito make me think the main worry would then be it being badly *implemented*

all you need to know about UEFI is 1) firmware OEMs are trash, and 2) Microsoft

Sompi: in the absence of proper advice on non-sse2 IA-32, I can only think of mentioning linux from scratch and gentoo - that is, check if their instructions or scripts have something for that

polticial off-topic: never been to the US, no right to vote, but it would be silly to say this matters only to the US, this has been spreading worldwide and I wonder how much of the current situation we can actually nail down to a few extremists in the US getting some more power.

Just like Stephen Falken, the rest of the world could try to stay away from this in an island, but would certainly not be shielded from Global Thermonuclear War.

gmail has 20 megabytes of javascript

I remember when it had zero

and worked better

I wonder how much does the Mastodon web UI have, it feels heavy (and isn't really well designed UI-wise, and also doesn't linkify news: anymore)

now, when both landlines and the 3G network was closed down, most places here have only 2G, so 256 kbps GPRS is the best connection possible

6 MB of javascript takes a minute to load even with a 1 Mbps connection

putititonelinewithoutspaces!thatsurewillmakeitgofaster!

Another common thing with that javascript crap that in many cases those scripts just somehow break when things load too slowly

So even when the scripts eventually load, they don't work

it tells something when it's obvious that not even the smallest possible deviations in the setup were considered in testing

say, when CloudFlare deploys code that fails on the browsers it is inteded for and launches a DDoS on CloudFlare

Discord has 21 MB of JavaScript

This has 34 kB: sininenankka.dy.fi:8082/index.htm

And works on Seamonkey

(the backend is written in C and a 266 MHz Pentium II is already very overkill for it)

Discord is... [censored]. Gave up on even giving it a try when to run it again I was presented with a new EULA, on top of bad usability.

blue duck, is that a reference to something?

most modern websites are literally unusable, not only with other browsers than chrome or firefox, but also with less than 10 Mbps internet connection

njsg: just a random thing

even facebook has ~always been like that - the scripts just "give up" when the page loads too slowly. it requires a relatively wide bandwidth connection to even be able to use the chat there

I have 25Mbps here, and even with that the wonders of modern Internet routing often ensures me a miserable load time experience for many websites, when I get sent through the slowest/most congested possible route

that chat thing that I made works just fine with a GPRS connection. I once tried to use the Discord phone spyware app with a GPRS connection and it could not receive and send the text messages in the chat, apparently there was too much overhead in them

You don't need GPRS to experience crap mobile service

here in Venezuela we have unusable 3G and unusable 4G, where sometimes even opening something as simple as a static page will stall harder than a 737 MAX

I've seen decent mobile data over 3G, but that may have been because I didn't wander too far from densely-populated areas, and I didn't try 20MB JS webapps

tomman: don't worry, the 737 MAX comes with an automatic system to help avoiding some stalls

Finland demolished its landline network and now 3G was also closed. Now we have 4G and 5G but theis coverage is very limited, and 2G that works almost everywhere but it is far too slow for today's needs. And also the 2G is going down in 2029

s/theis/their/

elisa ended the cheaper "lower-bandwidth" data package for prepaids, I was expecting that to mean they at least had much more bandwidth to offer now?

DATA6KK or what it was

Also the mobile frequencies are very crowded in urban areas. Now the problem is even worse when the landlines don't exist anymore, because now everyone has to use mobile

2G is essentially dead here, except for the state telco that keeps it going (the sole reason you can still buy $12 2G-only dumbphones on stores)

They started selling 5G connections with dedicated frequencies

it feels like we could be exploiting lower-bandwidth mobile data connectivity a lot more, but now if what we start seeing are these dozen MB webapps, then I have no idea? Fiber to the handset?

the other two telcos don't allow you to roam on 2G anymore

but on the other side, they have no desire at all to roll out VoLTE, and 5G is on the drawing board yet here

This current direction of development is just stupid. Only buzzwords everywhere and no-one cares if the things actually work or not

(supposedly this year will be the spectrum auction)

and meanwhile at least in .pt terrestrial TV took several blows from the new mobile radio frequencies

in the meanwhile, placing a phone call that won't get dropped after 3 seconds or expecting a SMS to be delivered on time is like playing lottery

but then telcos assume everybody uses WhatsApp anyway :/

Mobile networks are not scalable. We cannot just create new frequencies like we can build new phonelines or fibers

first a shutdown of PAL for more frequencies, then a change because 5G needs other frequencies

The analog shutdown is not happening here anytime soon, but analog TV barely survives here anyway

..and so does digital TV

nobody watches OTA TV in Venezuela, it's pure poison

(the local DTV transmitter in my city has been dead since 2021 or so)

tomman: shutdown in .pt was done so badly - mostly because IIRC there was a screw up in DVB-T signal coverage, the main thing it achieved was pushing some people towards "special" plans created by cable TV operators.

Everybody here has cable or satellite anyway

nobody watches OTA TV unless you like political propaganda or stale telenovelas

now, some cablecos are shutting down coax for fiber, so that means no more freeloading for those that were moochin' free cable TV

reportedly while DVB-T should be more resilient to signal loss, they just took advantage of that to save on the coverage. No idea if this is true. But the DVB-T signal quality was crap when the switchoff happened.

We have glorious ISDB-Tb here, except that... we don't?

and unlike in ANY sane country, the government operates ALL the transmitters

a 100 Mbps VDSL2 connection was available here before the landlines were sold for scrap copper

tomman: hey, .pt actually has one operator which I think even fought to be able to retain their naming... they have fiber. You know what is "fiber"? coax.

it's a sort of hybrid between cable TV and OTA, where the government broadcasts a package of channels

"fiber" is apparently the product name

only 3 private networks were allowed to get DTV licenses

after that 50 Mbps 4G was the best what moneu could buy

and they have to broadcast through the government network, and in SD only

And the latency went 3x

regional/local channels? You're gotta get censored anyway

that same operator also manages to have less TV channels than some others doing EDFA over FTTH, despite they actually having the TV signal on the coax... encrypted or not it's there.

Now 5G is available but it costs _a lot_

And probably also kills insects

I have no hopes for 5G here

at best it will be as bad as 3G and 4G are here

Of course 5G is not as good as VDSL was

when you benchmark a 4G network and find that even your old pathetic 3Mbit ADSL beats the crap out of it, well...

Sompi: so now how's the broadband offering? wasn't that still a thing 10 years ago?

tomman: Latency is often more important than bandwidth

njsg: The landline network is demolished, so no broadband anymore

3Mbit, is that ADSL2+?

only mobile shit

Latency here is... well, let's say it eloquently: a passenger bus has better latency than our 3 4G networks _combined_7

njsg: worse, ADSL1

my DSLAM never got blessed :/

Sompi: so what, say, are the offerings to connect a house in PKS nowadays? fiber?

fortunately I managed to taste some glorious fiber since last October, albeit only 25Mbit of it

I don't know. Probably some apartment houses have fiber and some have only mobile.

and with CGNAT jail

tomman: I thought 2Mbit or above was at a different "level", so to speak

but hey, at least I no longer have outages 6 times a day!

and repair times measured in "years" or "how many Benjamins you have in your wallet?"

tomman: sadly I've mostly forgot ADSL lingo. Not sure if really "sadly", though

A friend moved to a relatively new house where there is no connection at all. The landlines were removed and the mobile tower is too close and its signal does not shine low enough

(the DSL still stays, $MOM refuses to get rid of it, and it works as a backup for the rare fiber outages)

trying to convince ISP support staff that "uh yeah you have a problem in your phone line, SNR is so low it's not manageable, would you just have it looked at" while they'd try to go over the script of requesting what the modem was, etc.

Over here they no longer care, but then it's the state telco

no, they had done something idiot to their line, and it was on their side, good thing I could disable negotiation and force something that'd get an usable SNR

reminds me when someone had misconfigured the upload on my DSLAM port

According to the official propaganda the connectivity is better than ever and only becomes better. In reality, most places, it is worse than it has been since the founding of the landline network

Sompi: that sounds like a nightmare, is this truly the wholly-connected, everywhere-connected world of the future!?

back then I was supposed to have 1M/512K, but my DSLAM went poof due to vandalism, and when they replaced it, they crippled the upload on mine

njsg: sininenankka.dy.fi/~sami/blogi/art.php?id=tietoverkot

so for 3 years (!!!) I had to endure 128K of upload... and constant refusals to get it checked because "the contract doesn't guarantee upload rates, only download"

eventually it took a BIGGER fault on my line to actually get a netops guy to check AND reset my DSLAM port properly!

Sompi: what's the future, mobile data will become worse than VR junaverkko?

tomman: I wish I could have gotten that kind of upgrade when I was on ADSL. I've quite disliked the "A" of ADSL

njsg: The future is that we have only 5G and 6G and their ranges are less than 100 m, so outside of cities there is no connection

Can You Hear Me Now?™

wait

kids these days don't do phone calls

it's all voice notes over WhatsCrapp (or other popular IM cellphone app that may not even work on computers)

in the cities there is mobile base stations everywhere and all birds and insects are dead and cancer rates are skyrocketing

that's the future

On one hand, I want to rewatch Dark Angel, OTOH, I fear it might resemble what you're describing *too much*.

they only use frequencies that don't go through living tissue, instead they release all their energy to any solid matter that they hit