Status meeting starts in 31 minutes - wiki.mozilla.org/SeaMonkey/StatusMeetings/2024-06-16

Status meeting starts in 22 minutes - wiki.mozilla.org/SeaMonkey/StatusMeetings/2024-06-16

checking in to say here for the meeting!

off to make some breakfast~

Status meeting time - wiki.mozilla.org/SeaMonkey/StatusMeetings/2024-06-16

Hi rsx11m

howdy!

doody?

exactly 8-)

hi rsx11m

hi .*

Who's taking minutes?

hi njsg

me

thanks

Nominees for Friends of the Fish Tank?

I nominate Myckel for providing 2.53 patches for the build system

seconded

2nded

oked

s@2nded@3rded@

Action Items

bau

so ctrl=alt=f shows a blank dark sreen

worn sindow

wrong window even

Status of the SeaMonkey Infrastructure

bau for me

ewong is slowly migrating linux boxes to Rocky from CentOS

makes sense...

meep

next one he's working on his aus2 (which is used for doing updates), so will have to wait for a few weeks after 2.53.19 gets pushed out

does this need a new update url or why the delay?

no will keep the existing update URL, just need to test the new box when it's not busy

so either taking over same ip or a DNS change only?

so no need to do anything different just the same rul will point to a different box

WG9s: DNS will just point to the new box

url

exactly

Status of the SeaMonkey Source Tree

frg is amazing? that is the status lol?

Needed a minor patch to keep comm-central builds working just moving things from convars.sh to moz.configure

all building I think. I finally pused the outstanding patches to central

but ou see i just added a new outstncing one.

My wip queue is now at the end of 67 with SpiderMonkey patches. Dynamic import and bigint working.

frg: how much would we loose if we attempted an esr68 spidermoney graft?

yay :D

Still not good becuase need to rebase the optional chaining and regexp stuff.

my all platforms builds using latest WIP Queue running now

as in swap it out

eh I'll look into that soonish, sorry to derail

ContributorTobin not possible. They removed add-on compartments and other stuff. I also left out some global properties and multi realm comparetment stuff. A

ah

Will look at some if it later. Also fallout fro stuff outside SpiderMonkey like static component registration.

if you grab the wip queue from gitlab.com/frg/seamonkey-wip check the series file.

Release Train

mozilla-central-push_421799.patch to mozilla-central-push_460238.patch is the so far taken out stuff.

added the 2.3.19 beta bugs and switched nighty builds to 2.53.20b1 pre

Sorry took a bit longer for personal reasons again.

I think we can do the beta tomorrow. Looking into some sec backports today.

is sm's nss set to default to dbm or sqlite

With Cent OS 7 basically EOl do we still want to do x86 builds? My Proposel would be to end this with 2.53.19

i think I asked this before but I forgot the answer

sqlite

dbm is still built for udpdates from old releases. Might end this too soon. We just need a watershed release then for automatic updates.

frg: I agree, on ending x86 build production but to avoid a huge backlash I'd keep at least functional x86 around.

so they can build em them selves

frg: good news, Rocky 8 still seems to build x86 fine

IanN: which gcc version you using?

ContributorTobin: gcc 8.5.0

good version

4 7 and 8 do REALLY well on mozilla code in my experience

10 isn't too bad either

andit builds fine using clang for me

Personally I don''t want to rip out the code and WG9s can still build x86 for people who need it but it is a waste imho.

frg: anyway re nss so all that is left is to what shut down the migration?

but I am using clang 15.0.5 had an issue using later versions with it not bing able to find X11 libraries becuase of some build config issue

not sure what current state of that is have not tried later clang versions later

For the dbm issue I just remeber that this would just need a readme section. 2.49.x and older are not set for automatic updates. So can take it out and put it in the release notes to update to an older version forst if you update from pre 2.53

ment had not tried recenlty

Anyway dbm maybe for 2.53.20

frg: as much as *I* like dbm for sustainability (unless you wanna fork nss) it just needs to go away as one less complication.

2.53.19b1 has lots of stuff in but not much for eb compatibility. More a groundwork release.

yeah speaking of tthat fedora's start page is broken in sm lol

so please test. It is super stable for me.

I'll run the queue and thrash for y'all

ContributorTobin: broken in sm or server-side?

I'm getting a "... is offline" message

Timestamp: 6/16/24, 09:31:56 CDT

Error: SyntaxError: expected expression, got keyword 'import'

Source File: fedoraproject.org/_nuxt/entry.BoRD9q-g.js

Line: 23, Column: 24327

Source Code:

s:[],redirect:F==null?void 0:F.redirect,component:()=>r(()=>import("./404.DzcOTCdt.js"),__vite__mapDeps([0,1,2,3,4,5]),i

and some custom element placeholder animations

ah I was looking at fedora.org

just mentioning it cause well.. its set as the default homepage

version 2.53.18.2-fc39

late?

hi world anyway~

tomman: hi

ContributorTobin: that's probably default homepage for fedora builds, and if they have exquisite JS on the website their web site team needs to fix that. loads without JS for me, but seems heavy-ish?

tomman well i thought it was odd you left 5 mins before meting start to say off for bradkfast

and now I've got my breakfast~

Extensions Tracking

It needs dynamic imports. My wip build can't load it because it als need the new regexp.

Redirector remains broken because its bundled jQuery is getting blocked from doing some formerly allowed stuff

bau

haven't really looked in depth how to address that

but if the brower and the webpage are form fedora sounds like a fesora bug.

2.Next, Feature List, Planning and Roundtable

WG9s: I agree but doubt they will

bau

it's sad when there are FLOSS projects that use these pages that require select browsers, if not also powerful GPU acceleration. Almost makes it look like I'm trying to play some powerful 3D game...

Sync as usual has a weave 1.1 php and server solution but webcompat takes priority

and I have tycho weave client as well..

so same as alwys

no progress from me on the composer side, I didn't even manage to test a change to keep findbar search history between editors

I am thinking of still trying to get about:credits into 2.53.19 final i might try a all pages built-in no externl server required verison for now and then work on the website for 2.53.10

njsg: Doom 3 now runs on web browsers, jus' saying~

but does it run on seamonkey

now I have no authority to do this but if I did DOOM on SeaMonkey would be second only to general webcompat.. because DOOOOOOOM

there is also that link color issue that was mentioned here some weeks ago, I think that one might already have a suitable bug report. IIRC the issue was that a stylesheet is removing color customizations for anchors, and my question back then was why was that rule added, didn't check that yet

so page style is overriding accessibility override colors?

ContributorTobin: SeaMonkey can at least handle a lite version of DOOM, the DOM.

njsg: !

AOB

Wayland support. How far up the patchqueue is it cause widgit code is kinda modular so it would be a good canidate to toy with patches between the queue and release..

If i use the same approach as with SpiderMonkey about 5000 to 6000 patches.

so little ways off

well I'll put it on my list.. I'll also create a list. lol

aob bau

any other business? well, 2GiB RAM sure is not enough to have a few tabs open, funny how I manage to get SM killed even with NoScript enabled. No news there, and any handling from SeaMonkey itself would probably always require that I disable overcommit and the OOM killer.

Also, There is only XUL dot Org is no longer going to serve as an index and resource to mozilla projects due to lack of cross-community .. decency actually.. thus the xr subdomain will stand and the main domain is now redirecting to seamonkey..

okay, thanks for your time today. Next meeting is in 3 weeks time, same bat channel, same bat time

see you in July!

IanN: thanks

ContributorTobin use xr almost daily. Thanks for providing the service.

ContributorTobin: agreed

Oh yeah, github.com/binaryoutcast/neptune is where I am developing the new add-ons server software. Which will be mirrored to gitlab eventually

cu ... here

frg_Away: xr needs some love too

frg_Away: I dunno how mozilla gets along on dxr/searchfox alone

I wonder what happens with searchfox when mozilla switches to git. I still think doing this is their biggest error in the last years.

isn't all GiggityHub these days?

not yet

still, mastodon/fediverse might think it is because somebody picked that up a couple days ago

(again?)

searching on gitschlub even with a supported browser is painful

it was already a bit useless before they started requiring JS and these extra features

frg_Away: going git would be fine if they were capable of producing a full fidelity repo which they aren't but will botch.. OR they said yep hg is done we will host the hg archive forever and we are starting from a fresh commit in a new git repo

now... it's probably faster to clone and grep

with git, I wonder if they do have reasons to switch other than "git is shinier" and people hating non-git because it's non-git

though one thing mercurial doesn't have, I think, is shallow clones. But perhaps that could be added to mercurial somehow?

for the kind of insane bullshit this project pulls with hg it would be moronic to dump it.. but hg as a default for everything.. its overkill and git is faster and somewhat easier to use

I think they wnt to get rid of own infra and the people maintaining it.

CaptainTobin: the problem nowadays with git is that for 99% of the newcomers out there, "git == Microsoft Github"

true

tortoisehg is way better than tortoise git and available on linux and macOS too. It is a great gui shell and another reason to stay on hg.

it's like saying all copiers are Xerox

on windows .. sourcetree 1.5 .. on linux i am struggling

lol

"git is faster" sounds more like an invitation to improve mercurial or implement a different set of tools for mercurial repos. but well, it's their repos. As far as they do try to keep history... lack of history is one of the biggest headaches I have when trying to investigate code changes, and that sometimes happen when I hit the mozilla mercurial epoch :-D

I think hg and git have their pros and cons and neither can completely replace the others

one thing to take into account is that mercurial is a DVCS, so there's less of a point in comparing this to a migration from, say, CVS.

can anyone remember the command to check the libstdc or glibc requirements of a binary?

but back to Tobin's point, if you're on the "it's all github" group and use it mostly on the web, then that doesn't matter so much, or, if anything, is a change from DVCS to non-D VCS, because then the website will be a SPOF.

IanN: there was some elf dumping utility that'd list the symbol dependencies, and there one could grep for ...STDCXX with a version, sorry, don't remember more than that right now

well running a forge with mozilla sized code is a damned nightmare I do NOT recommend it..

is it ldd?

not sure, that lists dependencies on libraries, at least by default I'm not sure you'll get these symbols

I'm the captain now.

IanN: possibly readelf, it has a --symbols?

njsg: I think ldd -r -v

?

i think displays the same info just in different formats

thanks njsg and CaptainTobin

ignore no-csd

lol

from my output

I think ldd is also showing info for other files that are pulled in, now this might be what you want but might also show other stuff, say, if the current system requires a newer glibc++ from a non-seamonkey library, that'd not be a requirement for the build in another system? Or am I overthinking this?

ldd's man page mentions objdump too, I wonder how useful that is, here -p is showing "Version References:" near the end

man pages are sexist..

I read that on the internet

:D

ok, in that case I'll use the naming I most often use:

ldd's online manual page mentions [...]

online manual pages are abist, some people can't comprehend simple instructions or directions

ableist*

i read THAT on the internet too

sorry njsg lol

i like reading man pages on linux.die.net/man

so that'd make it an online online manual page?

linux die maaaaaan

lol

frg_Away: do you consider dictionaries to be a user submittable add-on or a project provided add-on

user submittable

locales?

CaptainTobin not sure if there will be any locales. I would say project until we actually encounter one in the wild for testing.

language packs are in webext format now and if soemthing like the old cZ pack would even work now beats me.

frg_Away: understood, I have updated the intended userlevels to be user, developer (user with a reviewed add-on), reviewer, staff (moderation escalation, recover stuff, delete stuff, and project add-ons like locales), and Admin who can override everything update anything delete anything.. its an Administrator lol

at first sigh I feel like suggesting "add-on developer" for clarity, but maybe that's excessive

(but well, if you're adding the descriptions nearby, then it of course isn't much of an issue anyway)

njsg: I just didn't feel like typing Add-ons repeatedly binaryoutcast/neptune #1

i updated the issue with em

njsg: I totally agree

the number of times I have said Add-ons this Add-ons that Add-ons Team Add-ons Site since 2014 is astronomical

I think my type/subtype system will work best cause there is no way of knowing if there won't be a whole NEW kind of extension tech but there always seem to be at MINIMUM.. extensions, themes, locales, and dicts so subtypes make a lot of sense

it has been years and I still am not used to the naming being "add-ons"

mozilla fucked this up when they made extension and add-on interchangable

heck, I'm from the era where having an extension manager was a new thing

when they introduced the overarching term

njsg: same..

in my view they aren't interchangeable, I've always seen add-on as a module that can be extension, theme, langpack, ... - but maybe that's misinterpretation?

it is the correct interpretation that I enforced on the moon

that reworking of What are Add-ons?

is the single reason anyone can know what the hell anyone is refering to when they complain something is busted

i was a dick about it tho

i would just reply with the add-ons manager text in a quote until they corrected them selves

best not to continue that behavior

njsg

remember Fig Newtons are not cookies they are fruit and cake

DuckDuckGo: search result descriptions may include traces of LLM.

or is this just randomness?

njsg: are there any remaining just... crawler search engines left

I'm not entirely sure what's going on there, somebody on Mastodon said it was using LLM-generated page descriptions, one thing for sure is that descriptions may change between refreshes

for some pages I see part of the content there, doesn't look far-fetched, but I didn't try further