why do I suspect github.com/binaryoutcast/mxp-predev…51efd611971c015af3ad0f48125893e2f52 are some of the reasons seamonkey builds but doesn't run on central

guess it doesn't matter mailnews at 122 is busted

of course I dunno if comm-release or comm-beta are even expected to compile so i guess unless I go back to the cureent esr and try all i can do at this code level is remove comm code while leaving comm repo support files and slot my example in

cause i need to understand what the difference is in building firefox vs something in comm cause it's all different now

CoordinatorTobin suite builds but might need some patches from gitlab.com/frg/seamonkey-central-patches

There is some stuff for later so that we don't need to rebase almost every checkin like xul->xhtml and some half baked stuff.

TB should build ok.

yeah they finally got back to me and said the same thing.. I dunno i think studing it from a known working configuration would be better than trying to pin the tail on the m-c patch

well for suite without replacement custom elements to take up for the bindings it won't function but i did want to see if i could get anything out of it

also to test the changes I made to the build system because suite cannot be mach run for the same reason my example couldn't..

oh i am totally gonna apply those

for great jusice

or something

CaptainTobin: you seemd intereste in shy seamonkey builds under comm-central but it does not run. i am responsible for one half of that

It builds becuase i ws asked to do nightly builds to make sure it builds but was told no interest on if it runs.

as far as i know there are no plans to ever base anythong on any mozilla version expecpt maybe esr68

is talk of going back to a 2.57 base on esr60 code. but we really don;t want to lose the old extensions

the idea of making sure it builds is to try to avoid major issues if need ot bakcport a security fix in the mozilla code

would alredy have a patch to fix any comm-central change reqauired

anyway your post prompted me to do what i only do occasionly is to test my build to make sure it at least starts and found that current wource did not becuase of missing a library in the packaging step so i at least fixed that

so all we try to do with central is it vuilds and if you run it it will not get a statup error or hang in a way preventing you from exiting easily

CaptainTobin: so the fact that it builds is my fault. the fact that they don;t run is a decision made abve my pay grade.

I am not seriously considering using suite on mainline mozilla or suggesting that is what SeaMonkey should do.. but suite in c-c IS another point of reference one I am pretty familar with by now lol

i mean one COULD

so only effort on seamonkey team is to keep it building without errors.

it wouldn't end up identical as now but it could be close but mailnews and composer both make that way too much for one person to attempt

i have less faith in mozilla caring aobut us now the mitchel is gone.

I don't NEED mozilla to care very much, that's the idea but they should care .. a smidgin.

only person left in mozilla upper management who i conidered a friend

and hopefully accept a few patches so *I* can care instead lol

and now even she is gone

yeah

if you have patches you can submit and sk for either frg or iann ro review

i may, i say i don't want to just do netscape but even if i do other stuff EVENTUALLY i am gonnd do netscape.. it's in my soul.

was just a we have enough to do fixing 2.53 nd no time to work on current trunk

No i know dude

we might ebven have a list of things that need to be badkported someplace iknow there is a bug with some of them

once I have a wrangle on modern mozilla i plan to contribute more to seamonkey beyond what is still pending

but i think at some point even this bug was not being updated anymore

Well the way I am organizing BinOC/Labs is into a projects category and an experiments category and no matter how much I resist it I know I will eventually make suite an experiment even as I create stuff from close-to-scratch

are you seriously considering a jump to 68?

not now just after geting sebextensions to work and esr60 ws thought that was next step but also though proboably never go past that

issue had todo with geting to trunk mans breaking reasons people sill love semonkey and don;t want to run firefox so would need to be a not just make it work like firefox but still be better for at least some people

Now if I were gonna help coordinate a project for suite on current mozilla i'd take the following steps.. make mailnews and composer build optional .. because they bloody well should be and also so one can focus on one thing at a time. Navigator, mailnews, then composer. With "communicator" common bits updated as required.

otherqise could go back to an old idea i had forget the suite as a seperate thig expect a wrapper thing that lets you run firefox, thnderbird and chatxilla inthe same window and looks like a suite thing

and sets up defualt config options in ech app a defual to run in the app that is part of this psuedo suite

if you merged firefox into thunderbird you'd have chatcore so no need for chatzilla or even better a chatzilla ui using chatcore

and it would still be technically communicator based

cause Thunderbird is a direct decendant where Firefox is not of Navigator

could maybe use a different browser than firefox for the browser part of this

The best bet is to turn Thunderbird into the suite

aside from .. from scratch-ish

could maybe even add a say to use a different chat or mail app in this psuedo suite thing

that would be the closest way to retain familarity in dev and end-user wise (after some ux overhauls)

You know I had this idea before I joined Pale Moon.

so I have had over a decade to consider how it could be done

back when firefox and tb fist came out i susgested this and no one thought it was even woth considering

well at the time it was a harder sell cause xpfe transition worked.

wait

sorry that was later

but it still worked

WG9s: You recall the orginal idea was there would be a common xulrunner binary on the system and firefox thunderbird seamonkey sunbird and everyone else would be xulapps on that runtime

only thing i run on hat is chatzilla

I wish ascrod hadn't burned out his fork as a standalone was pretty good

and that is more becuase back in the day when i did more work with mozilla on firefox it was a pain-in-the-ass to run it as an extension becuase if i was trying to work with someone else on a fix everythime i wanted to test a newer version i had to lose contact with the person i was colaborating with

There is one thing tho WG9s .. waterfox has technology some of it from earlier thunderbird that does bootstrap "xul" extensions and can emulate overlays and read chrome manifest and this stuff seemingly works on mozilla-current.. so even the xul-extension capability question HAS a resolution atm anyway

Themes tho are kinda fucked lol

but i also remeber back around 1980 toimeframe when i got my first personal computer there was a company called peachtree software and the had a word processor and ithink three other apps and decided to sell them as a suite and all they did was write a grapical front end thing that basiclly asked wich one of these 4 stand alone apps do you want to run. people seemed to think this was integration

Well the Internet Explorer "suite" was standalone programs (save for mshtml)

I'd hate for seamonkey to cheap out on that solution tho..

soyou could run the suite but that was when these things had 64 meg of memory only so you were wateing some of that just to run the which app do you want tun thing

I wish mailnews external linkage was completed (and binary xpcom components rendered largely worthless)

if you have stuff you thing is adding wothy submit a bug and cc me and i can add the patch to my builds and if you get frg or ianN to approve will land on comm-central

then thunderbird could conform to the old xulrunner concept

cause mailnews would be shipped with the application part not rolled into the platform part

i have no idea if our gcurrent gitlab queue for central is public if is not you ned to talk to frg about that

otherwise you canuse the zip version on my website which is public

The fun bit comes when I satisfied my big job overdue promses and go searching and doing all the little things lol

I seriously think SeaMonkey needs a UX Polish pass nothing radical just well like some of the Borealis UX fixups I already gave frg

Just the me sitting there finding all the little inconsistancies and making them consistant

be a good task for when I need to clear my head in the future

was that a polish or a polish patch/ there is a big difference

what did I type?

polish would mean only has to do with the locale in poland

you typed polish

the other one lol

Polish

pol·ish

verb

make the surface of (something) smooth and shiny by rubbing it.

sorry my lef hand shift key on this keyboard does not work but i can't type trying to use it

that would be small p

that is what i thought you meant

my left index figure hurts like hell and typing isn't always easy

it is one of those odd words that somehow if you put it as first word of the sentence becomes ambiguous

dunno what I did to it but the tendon on the palm side seems like it's overextended

becuase the only diffence is the capitailization

capital p refers to the country

i never thought how it is the same spelling

weird

not that it is that I didn't think about it

but then if first work in sentence would be capital p either way

would have been better if one of them had an extra l in it

like pollised for make it shiny

pollish

but yeah WG9s this is what I have had to do to get arbitrary building and decided comm was a good target

I am having enough trouble trying to figure out how to do builds uing frg's wip patch queue and getting my back door fixed because it is a condo so have to deal with condo assholes as well as contractors

these?

no he has a wip queue for things he does not consider ready to put in 2.53 but i have better faciities for doing linu and macos builds so i think i finally got it so i can build and post linux and macos builds that he can test and then maybe we can get this stuff into 2.53.19b1

ah

is a mess of js stuff that i am not sure what it fixes and some python3 fixes

Well I am out of the windows game like completely now VMWare is too much trouble and so is virtualzing windows

but i can test on linux

of course

i am not doing windows builds becuse that is his deveolpemt env so before he tells me to do test builds he has already done tests there

i have no facility to test macos builds but he does so kind of an odd arrangement

me neither

Not odd

not to me

i can give you a link to where the test budls are but would be better if i could find out what is supposed to be new so you would know what to test i am only building not testing relying on frg to do testing and iann for linux testing and thyey goth know what the new features are supposed to be but i have no clue for this particular effor my involvement is only in doing builds

but is odd i have the build environment and he actually has physical hardware and vms to test on

I am unfamilar with mozilla testing framework but I am planning to learn it cause well.. it's there and has to be dealt with.

and i do not i am just doing cross compiles

i have neither physical hardware nor a vm

nor an apple license

This isn't that different than the situation UXP and co fell into.. Moonchild did windows builds and testing Travis did Linux Builds and testing and I did both as needed but primarily windows

well i dunno about today

but for a while all you needed was the apple efi .. key id thing and inject that into a vm using kvm and then you can install macos on it

unless amd

CaptainTobin: i consider the way mozilla wants testing patches landed is lame if you are doing something new then it makes sense you should provide new features is fine, but if a bug then there should be a test already that fails and is expected to pass if your patch works written by qa and then you need to post a fix that makes the test pass without braking any other patches

I had planned to get it setup even had access to hardware i could extract an id from but the ones who have done it or know the process wouldn't help me do it

I am not a fan of mozilla's method of testing I prefer real world testing bashing it to see if it breaks cause THAT is how it will break in the field not by the conditions of an obsolete test case that needs rewritten

fixing a bug and wiring a viald test are really 2 separate disciplines and expecting someone to be able to do both effectively is kind of dreaming. any idiot can write what they claim is a vlaid test that will say their patch works as epected even if it really tests nothing.

and the reviewer has probably less idea of how the test environment works than the patch writer works

and then we have the always present if you r= my patch i will r= yours syndrome

Well I don't HAVE to write tests for my own shit

but maybe i should

we currently have too many tests. should be tests the responsibiltiy of the module owner that the module works as expected

You guys still doing the whole module ownership jazz lol

ah test for every bug fix checkin is really not the way to go the module test sould be fised to include testing if required

oh we don;t but i thought mozilla still did i thougt if you as for sr it went to the module owner

They likely do

but i dunno how strict they follow it these days

besides, Mozilla wants ALL "browsers" to have the same web capabilities

i am one of those idiots that would write a patch and ask for review by bz and pople would say he is jusch a hard marker why ask for him. my answer was he might be a hard marker but thenwahat i land eventually will probaly not be blamed for future issue plus i will learn more form his hard marking.

same

but then if all browser work the wame why would i use firefox/

CaptainTobin: all as in all? why? sounds like, dunno, wanting lynx to support webgl

So, even with all js and cookies enabled, github no longer works in seamonkey

i still find it odd that mozilla does not evnen know what they di that eventually resulted intheir loss of browser share

andr01d: are you surpised?

Not sure what the changes have been over the last month, but the spread of this phenomenon is making SM obsolete

andr01d: github seems to be playing with a IMHO not so brilliant design where now a bunch of things that used to just show require javascript, at the same time that they use very specific javascript features almost nobody supports

this is on github, really, they're the ones who made it impossible to, say, read a README or explore the repository file tree on the web

CaptainTobin, not a matter of surprise or not, it's a steady creep of web sites just not rendering in this browser

still: for commits or comparisons, append .diff; for files change blob to raw

for releases, there is a way to get the address of the HTML fragment of a specific release/tag, although I don't recall what precisely has to be replaced with "expanded_assets"

Ok.. overall seamonkey, lingering mcp-ers, anyone.. What is more important to you: Webcompat or xul extensions?

ws bak inthe day about 4 of us doing anything on thunderbird. scott mcgrggor 9good frient of ben goodger chief engineer on firefox) working on the front end dave moze working on the mailnews code me kind of working a bit on both and also testing the mime code 9a new thing at the time0 on different implemetations of mime and arvid doing the icons.

isn't this what its all about?

It's not just github either, there are more and more sites that just don't render. Historically this required enabling js and cookies, but now this doesn't work. There is some spreading web framework that is just not SM compatible

andr01d: it's the spreading of lack of error handling, and also of error handlers which blank the page. to the extent it probably is becoming more common that *disabling* javascript fixes sites

say, MDN

CaptainTobin, I'd really like to have xul extensions AND a browser that can render whatever this latest cancer is...

compat tables are dynamic, and those won't render, but the content, I'm still able to read it

andr01d: Github is a microsoft product and microsoft basically defacto controls non-google chromium on windows because they don't have mshtml or edgehtml anymore so they are gonna do like Microsoft always does and target it for THEIR browser only and maybe firefox

CaptainTobin: it's fun that Microsoft still has something for IE in exchange webmail :-)

it'll probably last until somebody at MS who does not like that finds out about it :-P

when microsoft started using wordpress I knew Microsoft was done as far as its own tech was conserned

this is just evil revenge as they eventually go down

they descided they wanted to get rid of tb and a reallyofficaily supported thing and came up with 3 choices for how we would like to proceed. well the 4 of use picked one of them ane they camd back with a tough shit tyou picked the wrong one we are doing what we want anyeay and scott got pissed and quit and then ben subsequently quit went to google and that is there the chromium project and...

...google chrome came from this would have been the next gen mozilla browser had they not pissed all of us off over thunderbird

njsg, I always run no js as default, and only enable as required. But github is just totally borken now

CaptainTobin: oh? what about the time they tried to mix NT and non-NT windows? although I guess accounts differ on how stable the result actually was.

What.. ME?

lol

WG9s: got a linux mozconfig handy?

doesn't have to be exact just something to start from

yes, ME

njsg: Win2k was superior

i wish Neptune had happened

instead of ME

would have been far better save for dos gamers

CaptainTobin: I won't argue with that, it looks like it'd (NT 5.0) be superior, although I've never used it personally

njsg: You want consistancy? You want Win2k

I've used NT 5.1 and NT 6.1, and I can say 5.1 tends to be less stressing on me

even its more 95 holders fit in well with the refined aesthetic

holdovers*

wasn't it NT 5.0 that had UI design and colors actually picked with care?

that is where ME got its refined color scheme yes

NT 6.1 is... once in a while I'll remember something I'd do in Windows 4.10 and I find it has been removed. What was the last missing thing I found this way? Desktop toolbars?

I wanted to put a separate toolbar somewhere else with shortcuts and was about to create the draggable toolbar to then drag to the edge or otherwise see how would it be docked... and NT 6.1 does not have this.

njsg: gstatic.gvn360.com/2020/06/w2000.jpg

also, shift+restart in 4.10/98

but then I'd not expect that to be available in a different system as WinNT. But it was something interesting to see and use

WG9s: no module named six.moves

i think that is realted to the pending python3 fixes

andr01d On 90% of the web sites it is either dynamic imports or private class fields starting with #. I am on it. Dynamic imports is 60% done but like peeling an onion one patch needs 2 others.

it hink six is kind of depracated

well I can't build then

what version of python is this using?

cause python2 and older than 3.12 is crippled here

2 and 3.12 is bahhh... Not sure if 3.12 even works with central now.

not at all sure frg is more on top of tht working with someone else on that

it doesn't for PGO

and is flaky otherwise frg_Away

3.9 to 3.11 should work for 2.53 and central.

yeah except i have no six module for python 3.11

or any way to get it

because Mozilla central doesn't require it

AND fedora doesn't care

Are you on 8 or 9? 2.53 under Rocky 8 builds fine.

Fedora 39

buc might know.

i only have python 3.8 i have not tried building on f39 that is what i have onmy laptops but not on my build system

buc for some reason hates my guts even tho I never had any negative experience with him and the only specific one went positively

frg_Away, Thanks for the info! Looking forward to the dynamic import support!

WG9s: well it won't work casue the compat python3.11 fedora uses doesn't have pypi or six

because it is there just to build firefox

andr01d me too. But not in the next 2 weeks. Maybe in 4 if I am lucky. Pale Moon has it in but I don't want to do a shortcut and rebase endlessly afterwards.

I am tossing a coin right now if I should rebase or pull more stuff in.

rebase on what?

cause if you say UXP after I spent years trying to get you to do it I won't be very happy

Current wip queu queue wich is on Spidermonkey end of 62: gitlab.com/frg/seamonkey-wip

About 1400 more patches than current 2.53 :) Running it for days now and stable.

jesus

havy the dynamic import stuff up to end of 65 in but needs some work. Then basically only 2 bugs with 20 patches left to finish it.

so what can I do to build seamonkey

cause I do not have the six module or any other version of python available

Install EPEL 8 or ask buc

uhh

you know that makes no sense

I am on Fedora

not centos

and buc hates my guts for NO reason

Reason I only do VMs for building. Much easier.

Or ask Myckel the gentoo maintainer.

He did some work lately with python patches.

CaptainTobin: so you are trying to build 2.53 with the latest patches form gitlab/

I am trying to build release code first and foremost

on fedora 39

I can build UXP and I can build central

I cannot build seamonkey atm

oh so the release tar file/

no

from git

from on fedora 39

os wo trying to build the latest fedora release version i have no clue how to do that

yeah i don't think half those patches are even applicable and an rpm build != a regular build

i mean required

and none of them deal with python

yes just that Is all I know

CaptainTobin: you should perhaps ask stransky⊙rc

ask him what

he is kind of led on mozilla on redhat/fedora and works for redhat he probably could tell you how it is built becuase he probably did it.

WG9s if Firefox works probably not much help

yeah he will tell me to use the sourcerpm

and firefox works

oh i did not realize firefox worked

yes

so you are trying to build seamonkey using a version of mozilla code that we do not support building seamonkey with/

I can build UXP I can build mozilla-current day i cannot currently build seamonkey release from git because python3.11 has no pip nor package for python six

and without that

i can't even configure

no

i am trying to build current seamonkey release

cause that is the first thing I do

there is a pip3 package i believe

the default current version of python is 3.12

there is a 3.11-devel for firefox and thunderbird building

it does not have pip nor six specifically for it

I will get back to comm-central aftward

because obviously I can't make any assumptions

CaptainTobin: there is a python3-pip-23.2.1-1.fc39.noarch

based on either uxp or central

do you have that installed/

sigh

WG9s:

if that is install then pip should be there

and how do i make sure i get the python 3.11 and not the 3.12 version of the six module

and it is picked up by python3.11 not 3.12

that probablyh has more to do with where it installs that what version it is bet they are the same but in site packagesgets in a different subdir for each python version

htis is above my paygrade

there is no solution except refactoring six out of the codebase or provide it

frg_Away:

WG9s:

I don't get how this package on fedora is even building

it has no patched towards python

CaptainTobin that is where Myckel helps. wip also has more advanced build support already. I hope to move it over to regular 2.53 soon

sl are you suing the patches in gitlab for seamonkey-central-patches/

I am not using comm-central right now

this is gitlab.com/seamonkey-project/seamonkey-2.53-comm gitlab.com/seamonkey-project/seamonkey-2.53-mozilla @ 2_53_18_1_final

WG9s: hi sorry

this is gitlab.com/seamonkey-project/seamonkey-2.53-comm gitlab.com/seamonkey-project/seamonkey-2.53-mozilla @ 2_53_18_1_final

oh i thought you were trying to build using the fedora firefox source file

nah that's later

I need a baseline first

working code

to make sense of it all

ok i can probably get back to you on this but like i said i wont have time to work on it till saturday

Well I may try adding six to the python files in-tree

would be esier for me thogh to start with current 1.53.19b1pre as i always have that working

but not under f39

I don't want you to go to any extra work for me trying to figure this out

2.53.19b1pre

does b1pre even evaluate properly under vc?

cause I always thought it was +/-, a..., b... or pre

i can probvaly get 2.53.19b1pre working then try the same build scirpts form the 2.53.18-1-final tar source file

anyway irrelevant question

If you mean VS2022 no. needs ipc changes. Sucessfully built with latest ckang 17 under Windows but still use VS2019 mostly. Much more stable. VS2022 is the Windows 11 of compilers.

2015 did a good job back in the day

easier to start with the b1pre becuzse i have builds for that and log files from build and esyier to comapre this worked and this didn;t with the same source with a different fedora fersion

well nothing works cause python for me is missing crap lol

frg_Away: so you are saying good i never installed 2022/

WG9s works if you use clang

well good luck reproducing that env later with visual studio

there are already missing older version components

well you do know mozilla using clang for everything so no longer ests anythong using the msvc compiler

I was gonna build something needing 2017's compiler but now its missing some pieces

mozilla now boostraps tons of 2019 stuff. mach bloatstrap if you ask me.

it is bootstrapping 2022 now

i think

2017 should work for 2.53 but not tested lately.

i think if you want vs2022 stuff wen you run mach bootstrap you have to be using a .mozconfig that specifies building with vs2022

frg_Away: mozilla vendored six

that's how they manage it

try ac_add_options --with-visual-studio-version=2017 and set cc=cl

CaptainTobin which bug?

stand by

ah so mack bootstrap installs a sis module

somewhere in .mozbuild perhaps/

no vendored into the tree like autohell is now

and mapi headers

autoconf 2.13 is now in wip and mapi already in 2.53 latest release

err

... you should have gotten this by now

Main mozconfig including the last paste: paste.debian.net/1307456

I added mapi to interlink's source before I abandoned it as well

build shell mozconfig 3.4 paste.debian.net/1307457

i usually do an hg cone of mozilla-central and then step into that repo/mozilla and run ./mach bootstrap

Thanks I had forgotten about --with-visual-studio.. uxp development rendered that worthless by no one bothering to up vc and then by upping but not maintaining proper backwards compat when it was still possible

then things seem to work much better

oh and after you might want to do a dnf erase watchman

i hate watchman

CaptainTobin without it will pick the highets installed which is in your case probably 2022

seems to just slow things down and eat up cpu

I am avoiding running mach bloatstrap

if I can avoid it

Yeah disabled it hard under Windows too

btw i am totally keeping that

on windows i cba

cause it wasn't gonna use my default version either way

I recently put stuff in which picks clang if available. so export cc=cl is it for msvc

I on fedora 39 have to export the clang path because it's not picked up correctly by release

bootstarp wasn't so bad until 115. Now it is atrocious.

^boostrap

dpaste.org/rPCqx is the mozconfig i am using for mozilla-current

I haven't gotten as a far as creating the comm hacks yet but i have opened comm/ up to building and running anything

CaptainTobin: let me see where my sis module comes from perhaps i and tar up on directory inder .mozbuild so you don;t have to run the whole bootstrap

sleep time here so have fun

frg_Away: rest well

WG9s: it's cool man i appreciate it tho.. I think I am gonna port 1388340 and if that doesn't work then maybe I can ask you to go to extra special trouble for me ;)

I mean buc obviously has his own slightly unreproducable solutions I'd prefer it to simply work.

i am sure you can get behind that WG9s

this is way trickier thatn i thought you might have to do a mach bloatstrap

are we talking sm release or c-c?

i usually run mach bootstrap on current mozilla-central

ah i got around it by including UXP deps plus deps from the firefox spec file

on linux

i flat gave up on windows

if you do a cd `/.mozbuild and then do a find . -name six.py -print what do you get/

cd ~/.mozbuild like i said my left had shift key does not work. thank you so much my cat bella for spilling wine on my keyboard

he left had shift and esc key do nothing now

there is nothing there for seamonkey release

but my mozilla-release shit is there

in you .mozbuild with that find ?

or you did siomething different

I will not run mach bootstrap on linux

i am trying to avoid that

the problem is mac is gois going to set it up so that it will look for something in .mozbuild whifh is what the mach bootstrap that you did not run would have saved

I guess I dunno what you are asking me for

right now I want to build the current release of seamonkey from gitlab

then I want to build comm-central

well i can tar up a subdir of .mozbuild so you can try untarring that

looks cleaner and longer term ;)

and a lot less file trasfer

those directories get large

the thing with mach bootstrap is it does a much of crap it installs a mess of packages that you would probably need to do builds of anything else anyway and the rest of the crap all coes in .mozbuild so is a bunch of crap but only imacts mozaill code builds so if you have the space is is fine the only thing i refuse to install is watchman but you can always uninstall it after i also usually...

...put a dnf in /usr/local/bin that looks like this

#!/bin/bash

if [ "$1" != "install" -o "$2" != "watchman" ]

then

/usr/bin/dnf $* || exit 1

else

echo "WARNING: Ignoring 'dnf install watchman'" >&2

fi

exit 0

six is in the seamonkey tree

is it hooked up?

this only works after the intial mach bootstrap so if it later decides wants to add bootstrap but the orgianl watchman is in the midst of a mess of other crap more difficult to do. but once you do intial mach bootstrap if you erase watchman than subsequwent wones will not try to install it ever this is the only thing it isnstalls that i think i don;t want

wtf

I seriously do not understand this

seamonkey has the six package vendored

like i said the rest of the bolot goes into .mozbuild so only takes up disk space does not slow down the sytem.

and it is listed in buikd/vertualenv_packages

that is just my opinion, i could be wrong

i can't screw this system up

so i dunno

why isn't it working

hmm ok i have it in ~/.mozbuild/version-control-tools/third_party/python/urllib3/urllib3/packages/six.py

yes it is there as well

but the canoncial location is third_party/python/six

as a vendored module

obviously this work is incomplete

but thn i have never tried to build with this system why i said maybe i will get back to you on friday with an idea

because it fails so early it doesn't do anything except have the error about six

HEY WG9s

guess what happens when I run mach bootstrap on seamonkey release

mo module named six.moves

i can't bootstrap

on sm release

try just doing a hg clone;//hg.mozilla.org/mozilla-central and cd to mozilla-central/mozilla and then ./mach bootstrap

ok why?

how will this help me build seamonkey from gitlab

but after i would do a hg erase watchman

and then kill any running processes named watchman

i also don't use hg

i use git

so like ps-ef \ grep watchman and then klill things that look like running watchman

or just reboot

how will this affect my system configuration?

might install some additional -dev packages

-devel

i mean

should not break anything

you know fedora doesn't do that

run bootstrap

or slow down anything

i do my builds on fedora

but there is no way the python deps are gonna match between virtualenvs

just not on fedora39

which version then?

it should on builds rebuild the virtualenv i have found doing 2 builds simultaneously is prblematic but that seems more rust than pyton related

i no longer attemtp to do linux-64 and macos cross compile at the same time as i decided that fixed multiple issues but now can;t do it anyway becuase they require 2 different rust versions

I could try linux-63 and linux-32 at same time but based on my previous experince better to use a bigger number on the mak -j thing than to run 2 builds at once

i am trying a fresh mozilla-central clone and mzch bootstrap to see what i get from the find ~/.mozboot -name sys.py -print

six.py not sys

ind ~/.mozbuild -name six.py -print

there is nothing there

WG9s:

i am not doing central right now

i am doing the release code

2.53.18.1

i knw but i find the mach bootstrap in central is the one that seems to allow both

because i do both

also seamonkey stores its virtualenv in the object dir not homedir

and virtualenvs are pathlocked

and cannot be moved

maybe i am just confusing my self

i dunno

but i seem to also get six.py in ~/.mozbuild/version-control-tools/third_party/python

yes they vendored it to third_party and then reduced the duplication

later

seamonkey has at least four versions

in the tree

and none of them are used

well here is one thing

this is calling python3

directly from mach

that will be python 3.12

on here

not the python3.11 i invoked it with

but .mozbuild seems to be 16gig of disk space

someone fucked up

epony got klined

lthough i have extra because i have mutiple clang versions pus tar files for each version so mine is bigger than normal

... i shouldn't ACTUALLY be surprised

that I would have to treat sm2.53 more like UXP than modern mozilla

suppose I could do a chroot

will el9 work cause I could chroot install el9

WG9s: a small fc38 chroot would do the trick it has 3.11

for the minute I suspect

well tahat could work also

sorry i amabout ready to quit for the day and busy tomorrow which is why i said maybe i can duplicate your problem and work on it saturday

WG9s: Well I will try the chroot route and and if you get around to it I'll be around on saturday

also build central

for fun

and report

as long as you don't expect it to work and add all the patches in gitlab.com/frg/seamonkey-central-patches

it will not build without them

yeah

that is easy enough

i say theoretically

lol

if you don;t have access 9not sure this one is public0 patches also at wg9s.com/comm-central/patches/seamonkey-central-patches.zip

I have access to that repo

and i am not logged in

so public

ok guess loos like will buld without these just will not pass mach package step i think but beter to build with it becuase that is what works for me

Well I'll see and then apply the patches

just figure better if you have same source otherwise you saying dowes not work for you and me saying well it works for me is we are not testing the same thing

better if we both test the same thing otherwise we might be chasing our tails around forever

No I agree

but building without and building with will be valuable to me

beyond just reading the patches

that is fine when we get to trying to figure out if your builds fail and mine pass at that point better to be using the same source

also btw i post what compilers and mis for comm-central in hg.mozilla.org/comm-central/seamonkey-info.txt

also btw i post what compilers and mis for comm-central in hg9s.com/comm-central/seamonkey-info.txt

okk one more try htps;//www.wg9s.com/comm-central/seamonkey-info.txt

wow i need to get some sleep i think

have a good night

WG9s: rest well