phoronix.com/news/Chrome-Deprecating-JPEG-XL the push for WebM is strong, it seems

tomman We have not even implemented it. 110 will be the last version for 7 and 8.1 too. No reason to follow this also.

frg_Away: oh wow, Chrome is giving the kiss of death to W7...

many public offices I know here won't be pleased... but then, many of them are still stuck on ol' Compaq Evo pizzaboxes running XP, so they may not even care

Status meeting starts in 30 minutes - wiki.mozilla.org/SeaMonkey/StatusMeetings/2022-10-30

Still running it with esu. 8.1 too. Seems I need to get 2 more server 2019 versions if MS does not extend 7. But not for Chrome. Can go to the dumpster for all I care.

did windows NT 6.1's share within the overall "universe" of windows installs decrease *that* much?

many FOSS projects are dumping or in the process of deprecating W7

(and W8.x)

how does that overlap with projects pushing, say, discourse? :-P

for example a certain popular GameCube emulator ripped out W7/8 support a few months ago

njsg no still 10% I think and more then Win 11 garbage.

mainly because of the move to Qt 6, but also because of newest DX/Vulkan versions

noone cared about W8.x, but the bloodshed on W7 communities will be brutal

now are projects in general doing that because some compiler solution will stop supporting windows NT6.1, or something like that?

I wonder if Chrome dropping W7 also means Steam will also kill W7 support soon

(since it relies on CEF)

VBox 7 is a dog now too because of gfx. One test install and that is all for now.

many games on sale on Steam don't even run properly on W10/11 and have been largely abandoned by its publishers

sure, there is Proton... which is rubbish for anything made in Japan, or using anticheat junk

and, heh, windows-wise, I think I don't have licenses for anything newer than NT6.1

njsg: Jeff Atwood and its minions would just tell you to buy a new cellphone or an ARMac, of course

and by "new cellphone", they mean a iPhone, naturally. At worst, a flagship running Android 13

I do have two legit W10 licenses here, they're called "install W10 with your W7 key"

why would I buy a cellphone from a manufacturer whose quality control apparently doesn't cover "working GSM radio"?

because you're expected to live next to a test 5G NR radiobase, or always tethered to gigabit WLAN

fun fact: have a 4G phone, just upgraded the SIM card from my ol' KrapOS junk to one of those triple-cut 4G USIMs

placed the SIM on my 4G phone, and it refuses to connect to the 4G network

it will be oh-so-wonderful to call Movistar and waste a day figuring out what they forgot to enable on their systems to enable 4G on my line

hi all here for the meeting

tomman: it refuses to connect to all networks, or just 4G?

the same phone does 4G just fine with a freshly activated 4G SIM from another telco

just 4G, and the phone supports 4G on all the correct bands for my telcos here (a Redmi Note 11, telcos use bands 3 and 4 for LTE, which this phone clearly supports)

I've seen issues with cross-"generation" handover, not sure if with the card or with the handset

the phone can see the Movistar 4G network, but the BTS just refuses to let my phone register - in 3G it works fine

(on android)

KrapOS, is that a fork of KaiOS?

njsg Windows 10/11 still activates with a 6.1 one. For 8.1 I bought cheap update packages for under 20 Euros. Server is a bit more expensive so a bit hard to find legit ones cheap.

frg: Microsoft had some very obscure program for combating piracy in Latam: 10 years ago you could buy a W7 OEM key from authorized resellers for next to nothing

they never mentioned this program anywhere, but they offered cheap OEM keys (and slightly less cheaper retail keys)

all you got was a key, no COA sticker or setup media or anything

you could find random resellers on MercadoLibre

tomman now you just buy thinkpad frames or doors when you need one :)

frg: is it worth upgrading to 10/11, other than for stuff that won't support 6.1 anymore?

I mean, UI-wise, resource-usage-wise

it seems that program ended, or the deep discounts ended because all you can find are expensive Office 365 licenses, or equally expensive W10 licenses from the same resellers

Status meeting starts in 15 minutes - wiki.mozilla.org/SeaMonkey/StatusMeetings/2022-10-30

just key is how they operated with MSDNAA (that chicken-and-egg-problem service)

At some point in time W7/Office licenses were so cheap here that when people came with machines for reformatting them, I charged them the keys on the bill (they could save the Windows one if the machine had a valid COA)

surprisingly noone complained

11 is crap. 10 is crap but unless you don't want to run later stuff no choice. 10 with open shell menu is acceptable but you need to turn off a 100 things telemtery and unneeded services.

also make sure to have upgraded to an SSD

W10 is *horrible* on spinning rust, particularly newer releases

tomman well things like xbox integration and map updated and a 100 other crappy things need space and bandwith.

Status meeting starts in 5 minutes - wiki.mozilla.org/SeaMonkey/StatusMeetings/2022-10-30

Time for the Status meeting - wiki.mozilla.org/SeaMonkey/StatusMeetings/2022-10-30

hi IanN frg tomman njsg

hiya rsx11m

rsx11m Hi

hello rsx11m~

and hello world~!

hi all

on a rainy Sunday morning here

Who's taking minutes?

hi WG9s

me

thanks frg

Nominees for Friends of the Fish Tank?

hopefully sooner than the last time

hi .*

I nominate WG9s for keeping central building

seconded and tripled

Well thank you!

quatered!

quartered even

fifthed

Action Items?

nothing from me

Status of the SeaMonkey Infrastructure?

.

Hi CaptainTobin

didn't mean to interrupt lol

hi CaptainTobin

We are having problems with macOS symbol generation. IanN is this fixed in the source or later Linux needed?

been working with ewong on backporting some changes needed for macOS symbols

is there something I can help with here? is this still an issue with my builds or just with the releases?

Don't forget to scan UXP's repo for mac stuff as nuke has been overtiming adding it all back and updating it since earlier this year.

With the releases I think.

frg: fixed by switching to using the dsymutil that comes with clang

was an issue with my builds earlier. had many things in log files saying this has no symbols but this appears to have gone away.

IanN great wasn't sure.

WG9s: I think you were already using a different dsymutil in your builds

Had no time lately to do builds

probably but using the one that comes in clang fixes I should probably fix my builds to do that as well

WG9s: have a look at build/macosx/cross-mozconfig.common

this is bug 1795820, right?

I think we need to switch to sdk 11 soon.

but then I can not get macOS cross compiles to build with a clang newer htan clan-11 perhaps I need to fix that issue first.

I believe most newer macos stuff is very suboptimal with apps using sdk less than 11

njsg yes and no. The crash is gone but the dsymutil complains about glibc symbols.

well i am using sdk 11.3 just using an older calng and I don;t remeber what the issue was with newer clang

you may want to consider dual sdk macos releases or just resign yourself to an era to cator to .. older or newer

my friends mother died and going to do memeorial sevice and funeral next week so kind of out of pocket until week after

How's ARMac support going, BTW?

This need to change at least. Wonder why it works with 11.3:

But maybe only needed for native builds and not cross ones.

i am familar with the 52 version of that file.. what are you wanting done to it?

One of the problems with Mozilla decideing to do Windows and Mac builds as cross compiles under Linux.

tomman might be possible but sdk 11 needed. The issue is notarization here. And I think we need to drop macOS 10.9 to 10.11 support. Can probably be shoestringed in but with no real mac developer around... SeaMonkey now crashes on Ventura.

right major versions

frg: trust me on this: drop 10.9 people will bitch but its fine.. but try and make a cursory effort to keep 10.11 or those mac users will freak

CaptainTobin Just a one liner from the later version: WaterfoxCo/Waterfox-Classic #162/commits/b19728259e740a6c501bc1c83f837b4b715517f9#diff-f1d079c082395fc05c928f28d41bf304d165fcbc575d6700bd40b691e572b362R55

njsg: bug 1417646, bug 1487603, bug 1492526, bug 1513798, bug 1515604, bug 1551690, bug 1357317, bug 1607193, bug 1678291, bug 1680152 and bug 1729611

Anything which can run 10.11 can run 10.12. I checked. And yes I know...

for another year imo

anything that can run windows 7 can run windows 10.. doesn't mean people want to lol

WG9s I think you use 11.3 only with central.

and the problem with Apple is that they love deprecating everything every year

wonder for how long they still intend to support x86

Till 10.14 I liked it. Then it did go downhill. But that is another thing.

let me see what sdk i use for what build

as soon as arm can do everything they want x86 mac will die a swift death

and a few years later ia32 and amd64 will die a mainstream death

Status of the SeaMonkey Source Tree?

All building. central still needs a few patches.

yes I am using 10.12 for 253

All is basically 2.53 and central :)

I am planning to update the git mozilla master branch ifIianN lets me.

update it to what?

to 1782740-NSPR4341-105a1.patch

frg: after 2.53.15 release?

IanN can do it today or after release. You are the boss.

Wonder if we should update comm too but still manageable. mozilla is too big again and takes ages to apply in full.

Oh for infra our cert expires in December so need to contact mcsmurf

frg: I would say after release

frg: but willing to have a discussion outside of the meeting

frg: which infra cert?

fine with me either way.

IanN signing cert for Windows binaries.

frg: ah, the one that isn't free...

second connection failure today. Either my provider or sun spots

frg_Away: mine seems to be stable today for a change ;-)

Now that I'm asking about ARM things, how about Windows/Linux ARM builds? Nobody really cares about Win/ARM, but nerds are convinced that Linux/ARM is the future, are we there yet?

frg: ah, the one that isn't free...

We might need a mac dev cert too if we need notarization. I think it is only $100 a year.

frg_Away: be glad that you're not stuck on government-owned DSL for life :D

If Mozilla really wanted to help open source and prove they aren't a just-us commie front like 90% of silicon valley, they'd offer authinticode signing services for free

How long do those certs last?

CaptainTobin: security researchers would quickly try to kill that since it would be "prone to abuse", Authenticode certs are on a layer of trust a bit higher than SSL certs

well hackers can afford them.. i can't

depends. I think we are on a 5 year plan and just need to regenarate.

again sorry.. being mildly disruptive again.

back :)

Release Train

I just did 2 security backpots which I need to test and then we are good for 2.53.15b1.

I think we have some good SM slang terms :P

yeah, I think we're just about done on l10n too

I am buying an r next time. As stated previously looking at the patches most no longer apply at all. All regression fixes for stuff not in. Does not mean we are scot free.

Pale Moon and co upped their gecko compat to 102 citing extreme ua bs yours according to the notes is set at 91.. you gonna up it for your next release or leave it?

Yes. l10n synched. IT (Italian) is now complete too again.

CaptainTobin 91 made the sites I visit happy. 102 would made no differenc because the others need later js features. So late in the game I do not want to update. Maybe for 2.53.16

Just figured I'd mention it cause it was a bullet point.

it's getting the balance correct

indeed

see what feedback we get with it at 91

they are a bit more advanced js feature wise in SOME areas while you been doing methodical backports

but understood.

later JS shinies are indeed a problem point - many sites are no longer breaking due to dumb UA sniffing, but because they want dynamic imports, named regex groups, emoji junk, or ol' WebComponents crapola

Extensions Tracking

for example, anything using SvelteJS

Yes need to fix up regexp and dynamic imports. Waterfox has them but at leats the regexp patches are hard to backport. buttercookie42 backported the import stuff and should be easier then.

UA sniffing is just icing on the cake

tomman: totally.. i think we are ALL aware of the handful of painful es features

Well as I have noted in here I do want to explore a js engine retrofit concept to help some of all of us to potentally leapfrog ahead dramatically but it remains only conceptual with one MINOR example of a poc when waterfox pulled the trick with 32's engine in esr31

not aware of any changes for extensions

nothing new for extensions. Palefill is at 1.23 and I already added it to the notes. I turne don webcomponents as a test even in my daily use build and for gitlab and github it mostly is all what is needed for browsing. So no add-on.

IanN your l10n backports might affect some extensions in 2.53.16 but the usual candidates seem to be ok.

I am likely gonna be maintaining a fork of that extension because reasons and it will be project agnostic or more properly not discriminate and try and accomidate for more than one target

I would still like a newer public pdf.js release

The add-on sdk is becoming a liability too.

tomman just add it to SeaMonkey. happy to take patches :)

dunno when the current one is gonna break, either due to a SM update, or some weird PDF out there wereaking havok

jetpack extensions are easily replaced by webextensions in most cases.. proper extensions not so much

frg: is the pdf.js in m-c fairly out of date?

rest of my installed addons are Just Fine™, even the hacked up ones

you won't be able to get rid of jetpack until devtools are free from it.. until then all you might do is modify xpinstall provider to refuse to install jetpack and just keep it for devtools support

pdfjs stuff.. may want to peak at what Basilisk is doing.. or the extension.. might offer some bits for you.

although it may be time to bring black Flashblock from the dead, get rid of the Flash bits, and turn it into a proper HTML5 video blocker

web ext support needs devs :) 2.53.16 might switch th webext langpacks and I need to check if we can do dictionaries too. When I looked firsta year ago or so too much missing.

maybe hacking that would be a nice learning experience

tomman all this need devs and they are not there. For the daily stuff it is mostly me and IanN. We have help for python now and the occasional contributions.

frg: your add-ons manager is already webex enabled.. making it accept dicts is dead easy cause that is already intercepted for bootstrap hackityhackhackjob .. locales would be difficult but not impossible.. and no you don't need a million dependant mozpatches to do it.. just some familarity with the add-ons manager

I can look into it but likely not until after the first of the year

it accepting webext dicts isn't a problem, it does already

(as in, these get installed and show up in the add-ons manager)

but the dictionary won't actually work

I do know those components were updated maybe the format isn't backwards compatible

CaptainTobin yeah. I backported webext stuff and tried it with a local build Firefox 56.1x :) dictionaries came only in 61. I have the patche sin my queue and just need to check. Myckel did the webext l10n stuff and just a matter of putting it in.

UXP has been screwing with those libs too.. all stuff I have to port to my base as well

webext 1l10n is mostly build system and we are between 70 and 78 now for most parts. py3 next.

cause I was considering a hack to allow ME to accept moz-dicts

(and just mentioning just so that it shows up in logbot if anyone finds these lines with a search: IIRC you need to restart once between uninstalling the webext dictioary and installing the extension one.)

2.Next, Feature List, Planning and Roundtable

njsg: it might not know what to do with it

As usualy I try to get the whole chain in but a bit harder here because I don't want to "damage" classic extension support.

bau.

s@dictio@&n@

frg: Well I am likely the closest person to an expert on the add-ons manager around here so I will put poking at it on the list

there were some advancements i wanted to grab fron a later version to enhance the modified esr38 incarnation I use

same BAU for me too

BinOC XR has been updated last week to split the historical Cross-Reference mozilla and other trees to a different subdomain and m-c and sm-current are now indexed as comm under mozilla

I am still not 100% sure it is properly seperated but components/extensions seems to be webext. Would be too easy so we have stuff in browser too.

toolkit/mozapps/extensions is classic and some glue stuff.

frg: the add-ons manager is both

mozapps/extensions

was webex compramised from 41 on

before what would be called webex actors became a thing

but the add-ons manager is mutating it wasn't replaced like for fx4

AOB?

SWIFT release delayed so I am fine with the next meeting date.

Otherwise would have been a paid work weeked.

^weekend

meeting time will be different again in the US, I think?

frg: webextensions are a combo of a bunch of holes in the single thread security model allowing things that in xul would be security issues but in webex are features that use an abstraction later of hacks and "actor scripts" enabled by modified and corrupting the very systems that make any extension including webextensions possible

njsg: yes, the weekend after Halloween

and those actor scripts are both platform and app but have little to do with the add-ons manager repository or the xpiprovider

UTC still the same = 1500

rsx11m: thanks

frg: I am not just an angry tub of lard.. I am an angry tub of lard that knows the score ;)

thanks for everyone's time today, next meeting, same bat channel, same bat UTC

the meeting is at 3PM UK time is not at a specific UTC time

Oh one last thing

CaptainTobin This stuff is still over my head for sure. One fo the reasons I try to take backports unmodified if possible.

non-fxa infected weave is DEFF on the agenda!

IanN: full circle!

WG9s: we will be back to fixed UTC if and when UK gets rid of Summer Time ;-)

Well.. i am sure I was interesting and thought provoking of course but I hope I wasn't too disruptive tho

next meeting on 20th November

bye and happy vampire chasing!

rsx11m: happy ghost huntinh

rsx11m: happy ghost hunting

the only vampires I chase are in Touhou Project :D

booh! 8-)

are we mozcode vampires?

CaptainTobin: always

but watch out for the Gremlins

uhhhhuuuu

this is mozcode.. there never AREN'T gremlins

it is called GRE after all.. can't spell gremlins without GRE

now IanN wishes i was still maintaining my irredeemable asshole persona just so he wouldn't have had to hear that pun

and frg turns into its altar ego.

wasn't Gremlin an AMC car?

still here and away

frg_Away: tradition.

you don't even know WHY you honor that tradition yet you do anyway

i bet even IanN has forgotten

but he still does it

well if I am away just means I might not answer. If I am without _away I feel obliged to answer.

tomman: What's a Matador?

WG9s: everybody knows that the one true AMC is the Pacer, of course.

i miss the_bot

the only car that can make this sand-witch!

he always hated c++

oh yeah IanN frg_Away when i do get to the weave shit.. are we just giving up on jpake and easy 12 digit code setup because there isn't any kind of even semi-active impl for the keyserver and the python one was only updated to centos 6 capability by virtue of poking people who just only recently abandoned it.. the only person who knows how the keyserver works outside mozpersonnel is Moonchild and he isn't likely to help ANY of us regarding that..

However, if jpake isn't a consern then no problem. Besides, the whole 12 digit code was for cell phones anyway where cutting and pasting your recovery key wasn't feasable

not something I have looked at recently, I know ewong was looking at weave at one point

fsyncms however is a nearly 100% impl in php of the weave server and has been in operation at mcp for years without problem except storage resources of course

If we oparate a server it needs to be secure for sure. First step make it possible again to set up one locally and then see what goes. Not sure if azure is good here. Traffic might cost us an arm and a leg.

there are two things not easily doable vs when weave as active.. One is the jpake keyserver 12 digit code thing.. and two is the sync add-on installs.. not add-on prefs just the act of sync installing add-ons it knows about. I feel neither is that important cause no mobile .. I am not sure the version of olympia atbn has what it needs anymore anyway

frg_Away: main issue is io

and storage

space

those CAN be done in a vps.. but not very scailable

and even if you don't run your own server i know for a fact for some people being able to JUST hook into THEIR weave server is plenty

i can tell you that vs the python weave server and the php weave serevr it is no contest.. the php weave server is faster and uses less active resources

but then again.. seamonkey has what.. couple hundred thousand users? only about 10-20% at MOST would use sync and if you had a policy of stale purging say stale for 3 months then the data but not the account are cleared and a resync has to happen.. those are all successful strats that have been employed successfully without any budget to speak of

still moot until the component works for you again.

frg_Away: i wonder if i can make ABPrime work in SeaMonkey

for right now I need to integrate DOMi into my platform runtime.. i am sick of being without it

frg_Away: I have GOT to write a script that will shit out jar.mn

We gutted a few things. Not sure if latest ABP which worked is still compatible.

by taking a chrome package and scanning content locale and skin

frg_Away: ABPrime is based on 2.4 with enhancements from later versions but it is pre-bootstrap

EHPrime is EHH i used an xpcom init component to tell its bootstrap to operate so i could make it not restartless

:P

I really hate bootstrap because it is such a terrible hack

:P

I eventually want to re-merge the two extensions

cause EHH was spunoff from ABP

if you recall

ACTUALLY what I want to do is take gronehill's ubo classic engine and give it the ABP ux and call it ABOrigin.

yes. I did run ABP until it got desupported. Now happy with uBlock but ui is still lacking.

cause ubo's engine is just fine but its UX is horrible

well frg_Away maybe you can halp me at some point on it

I am considering splitting off all the third_party libs from my platform codebase into a submodule

half the damn codebase is non-mozilla code

how did that even happen

Well ChromeUtils.import backported but old methods still work. Ci. Cr. Cc. all work without declaring. Some legacy stuff gutted js versions iterators and generators.

861mg .. libs/.. 356mbs

frg_Away: I can do try blocks for the Components.import vs ChromeUtils.import

Both still work. ChromeUtils is faster if I believe the bug(s).

how is it faster.. as near as I can tell they merely renamed it

i don't see the benefit except shoving the word chrome everywhere

bypasses xpcom I think among other thisngs but has been some time since I did it.

ah well i don't want to encurage that lol

still maybe I will add an alias for Components as ChromeUtils then i can have patch compat without changing anything materally

sounds like a good compramise.. for me anyway

wouldn't you say frg_Away

call wise they operate the same right?

I can also check for my platform's runtimeVersion but that would need Components.* or ChromeUtils lol

CaptainTobin yes. I think stricter checks for the name. Must be unique because chached but this should be no problem here.

import is.. but this defineModuleGetter which for me is only named as part of integration.jsm as part of jsdownloads

defineModuleGetter seems to be basically a formalized way to assign a jsm to a specific symbol rather than its normal exported one

maybe the lazyGetter shit?

seems to be a replacement for XPCOMUtils.getLazyModuleGetter and calls to THAT are redirected to the chromeutils if it has all three args

so how do I get ChromeUtils.import to == Components.import and ChromeUtils.defineModuleGetter == XPCOMUtils.defineLazyModuleGetter

CaptainTobin the old calls still work so no sweat here

yeah i know but i was talking about for my tree.. how hard was it to do?

oh.. oh... oh

this dom component

no

i will not perpetuate the bypass of xpc for the likes of dom.

I'll just have to make the change by hand i guess

well ain't gonna come up too often.. js fe isn't something I want to stay heavily dependant on mozilla for

if I learn anything c/cpp it will be how to create xpcom components and the most BASIC core mozilla code and i will return some js components back to cpp

eithr by forward port and mod or by doing it my self..

well frg_Away i guess i have NO REASON NOT TO ADD IT..

chromeutils already existed it seems

so as long as I don't remove the older method there is no reason I can't use the newer

tho it will depend on the requirements of the action

not everything likely is a good idea to use this with my other conserns and priorities in play

doesn't look too bad

CaptainTobin Yes it was there. Just some added methods and other stuff.

well then i am LESS outraged.. less enough to do it but I don't have to like it :P

I did this very early if I remember.

well i suppose as long as i don't have to get rid of the full xpc involved process this speedhack and porting aid is sufficent

The rebased patches are here: gitlab.com/frg/seamonkey-253-patches/-/tree/master/mozilla-release.

checked-in is what is in master already.

do I need anything else other than the patches that touched chromeutils.cpp/h?

I mean other parts of 1431057 and 1391405 and 1412125

I can't remember. Now about 15000 patches and I only remember a few problems. I think this was mostly painless but you need to check the bugs for regressions and prerequisites.

I did the components stuff too at the time so that both Ci and Compnents.* worked but not a prerequisite if I remember right.

what do you mean?

oh

so wait are Ci and company always defined now?

yes

is that wise?

can't webcontent use it?

I presume it is only within the chrome context, but frg_Away possibly remembers

well I already found a blocker for me implimenting this.. it requires changes in xpconnect to.. i can't do that right now

Only to the extent it was available before. Interfaces are still marked Chromeonly and/or content and this goes for all the stuff. Unless mozilla screwed up no problem. And I took everything here too.

yeah but i don't want to be adding any alien modifications to xpc just yet not until i have done other research and if I get some messure or good success with spidermonkey retrofit i will either have to redo the work or automatcally gain part of it anyway either wya it will conflict for the time being

Well the old stuff still works and uBlock and NoScript using it too. So no problem either way. We ditched a few pieces ovet time whcih affected add-ons but only the ones giving us grief with porting and/or long deprecated..

it ain't a big deal for me to do right now but I should plan to do it

I need to audit all changes done to the js engine dom xpc and xpcom since esr52 so i have an idea what exactly needs done in order to meet or exceed my current status for my retrofit plan to have any hope of practical success

in my tree

i mean

imo for it to be beneficial I have to get at LEAST wholesale retrofit of esr68's spidermonkey MINIMUM

Got Bug 1410214 for dictionaries reapplied and only one imho unimportant piece still does not apply. But still no no go unfortunately. Something for 2.53.16 then. The dictinarty manifests look fishy without a version und the manifest parser dislikes them I think. Says add-on is corrupt.

frg_Away: in 52's add-ons manager webextensions install manifest support was a fuckin horrific hack that ONLY accepted webextensions as extensions.. i am not sure how much better 56 base was but i can imagine it didn't get any sort of proper cleanup started until xul extensions including MOST "system add-ons" were no longer xul

I am afraid to touch Bug 1389510 or everything would be easy. This one and the follow-ups is the biggest one giving me grief. But not sure what happens when I put it in.

CaptainTobin now on the way to 60 and much stuff in. Myckel already has language packs working.

I am fully willing to start some research and goal driven modifications to your manager to address your needs

it will just not be this year

but work toward it has already started cause i started it back in 2018 LOL

frg_Away: people like us have too much shit in our head BUT THE CHALLENGE TO USE THAT SHIT IS EVEN GREATER!

well i am not at all intrested in stylo so anything related to that or any rust code is non-applicable to me.. so best i can do is try and update dom and js

js the most right now

frg_Away: you said your js engine is roughly 60-ish with some select later patches right?

so what js engine level would you have to be able to attain wholesale to make it worthwhile to your efforts?

CaptainTobin still stuff missing for 60. Somewhere between mid 58 and 60 with parts up to 68.

so basically same deal to even be considered as an option absolute minimum retrofit would have to meet 68 or higher

else why bother

doing anything different*

would that be fair to say frg_Away

IanN_Away:

the dom changes will be extensive i imagine if you want full match but just to accomidate internal js functions it shouldn't be too insane .. just compiler which ain't gonna be a blocker for me for very long and skills I don't have but i can do everything up to those and then see where I am

I dopn't think a specific level is the happy one. We am just doing backports and add the patches.

Currently on Bug 1424420 and Bug 1424946 whcih uncorks some later stuff without rebasing the world.

yes but if i can make this viable you could leapfrog quite a bit i know you and everyone is skeptical but it could mean a lot

but i also know unless it can leapfrog significantly forward of where we both are then the cost/benefit is obvious

Yeah but still love to have the original patches rebased in. Makes tracking stuff so much easier. And thanks to mq we can reshuffle it later.

so if i came to you with a fully operational esr68 js engine running on your tree u'd turn it down ;)

if the answer is no then there is always hope!

i am interested in the changes you have done and taken in to mailnews core

depends if I can to apply later patches. If it is totally rebased yes unless we find a bunch of devs to maintain the fork. That is my biggest concern.

well the IDEA is nothing save build system build files are altered from the target wholesale retrofit.. after THAT whatever level it can't be brought over wholesale is the point you start patching and evolving up again and since it is FAR CLOSER you have more possibility to get even more recent shit while it might still matter before it gets replaced by something identical but incompatible es9384.2 stage 4

lol

dom-volution first stage is to stay with the js engine for minimum function and building THEN bring it up where you want it all meeting eventually at where str8 mozpatching continues on since you are doing it anyway this if it materalizes is a leapfrog event not so much an alt path least in how i envision you using the work

i see it if done in the way I think you would it would end up being a shortcut but nothing more

and anything I can make run on 52 should run no problem on your mozprogressed base

I am still hoping as well that those over at uxp would take advantage of such work i'd even let them control the js engine repo if only they would un-null route my server from mirroring from their git forge

frg_Away: this is where we know what the TRUE intent of this codebase is for.. the fallback in appshell for a content window is to load chrome://navigator/content/navigator.xul

the mozis can PRETEND browser.shitml is king but appshell says differently least up until they made appshell take a browser chrome uri by build var and provide NO FUCKIN FALLBACK

HEY frg.. can a jsm export a symbol that is the child of another symbol without overwriting the parent?

oh i see what they are doing

they are literally moving everything from Components.* to ChromeUtils in dom

yeah I actually don't want that BUT I will redefine the dom component and free up the symbol and create ChromeUtils.jsm which will access all the ChromeUtils shit for any scripts that need it

that seems like a solution i can live with

looks like on the JS side i can do whatever i want it is only used in like the push service and a few other places

CaptainTobin symbol export might depend on the scope. Totally unsure.

yeah i think I should just merge the two webidls name it MozChromeUtils and then create ChromeUtils.jsm as a compat shim

maybe even take over for some Components.* and XPCOMUtils.jsm functions

that sounds more my style

i might end up with it being filled out dom-wise in the end but i can not worry so much about it and move on for now

seem reasonable?

I gotta resolve these toolkit vs liberated lizard components

CaptainTobin ibb.co/pWH129X

Archive is 676 MB. I can upload it temporary but needs to go fast then.

well if you just wanna send a bare repo that's fine

or shove it up online somewhere

github gitlab code.binoc

Has the full comm-esr52 history. hggit did the job

amazing

i have never got hggit to work properly

of course i was doing it on m-c

or an m-r

that's just insane

i don't ACTUALLY have a public ftp dump ground for people

I should really add an account next time I regenerate it

CaptainTobin latest version is py3. Did it under thg which latest is also py3 and includes dulwitch

CaptainTobin can you grab it fast pinballz.net/custom/frg/52.zip

I WANT THIS LINE

WHATEVER LINE IT IS SERVING IT I WANT IT

i wish i could transfer from my server at 20 megabits

frg_Away: done btw

has two branches hg and master. hg can go. The latest two patches are just stuff I added. Was never uploaded so you might weant to cut them off.

now the fun bit.. forcing nginx to stay alive long enough and hope gogs processes fast enough to not timeout or be killed

these big ol repo pushes are FUN!

if i was SMART i would have extracted it on the server and import it locally but I arm not smert

Bug 1571287 also included. I think it was applied locally for 2.49.5 but not sure.

Windows Terminal.. the only GOOD thing to come out of microsoft in a fucking long time.. and it is a fuckin travisity the already existing need to build all of conhost should mean it could have ran on windows 7 and 8.. but nop fuk the terms for them

this really is the real deal.. wouldn't take this long otherwise

Yes even applied the merges: ibb.co/zFQ6WRx

compared the final repo content against my hg local one and identical. Same for the official branch and file differences are only because of the last 3 patches.

and now some sleep

resr well