autohide-slide

the feature is a good one tho.

I'd do it like this .. no its not in-code yet just mockup..

just a show when hovering over the field or when selected

The good thing when it is left you don't need to move the mouse so much. I am lazy and so like it better when left :)

nighty night for today

spend 20 minutes figuring out why dom inspector wasn't available.. cause I didn't have the configure flag you say.. and yes that is true but its more complex than that.. i worked on a patch to integrated DOMi as a regular suite component except that this queue doesn't have that patch and i don't have the configure option :P

It's cold here

compile seamonkey, warm the blasted space up

being a mozilla code developer is great in the winter!

in my experience, you want to compile webkit

well i would prefer not to put anymore load on this old system than it is normally used to it is getting up in years

ah, understandable

This PC is a battle computer that is what it was designed for .. as such it acts more like ME than any computer u'd pick up off a shelf.. My PC still has JUST enough old-pc in it to be an individual tempormental machine.. my electronic counterpart.

this system is LARGELY 2017-2018 era tho some newer some older .. it LIKES older Windows vs the Windows the hardware was designed for.. I figure this is the last machine that will have characteristics that differentiate it from someone elses similarly spec'd system

with AI-related garbage coming to microcode firmware and other bits of silicon.. non-conforming or quirky hardware won't be acceptable or the AI won't work

hmm

nsITobin hi

nsKITobin skylarke core 6th gen are the last good ones

upps ki :)

Client: HexChat 2.14.3 • OS: Rocky Linux release 8.10 (Green Obsidian) • CPU: AMD Ryzen 5 2600X Six-Core Processor (3.04GHz) • Memory: Physical: 30.5 GiB Total (25.7 GiB Free) Swap: 7.8 GiB Total (7.8 GiB Free) • Storage: 6.8 TB / 9.4 TB (2.6 TB Free) • VGA: Advanced Micro Devices, Inc. [AMD/ATI] Navi 23 [Radeon RX 6650 XT / 6700S / 6800S] @ Advanced Micro Devices, Inc. [AMD] Family 17h • Uptime: 19m 13s

too new :) No 8.1 server 2012 R2 support

wufuc normally solved that

the hardware has drivers

7 drivers might work but amd on 8.1 is totally unsupported

8.1 is unsupported now anyway ;)

so its all cool

depends. can still be updated like 7 till 2026 :)

8.1 worked well i just couldn't stand the UX

and cause win10 released all the hacks to make 8.1 died off pretty quick

make 8.1 better

8.1 feels faster. I seldom run it but has less bloat. With open shell menu it is good.

i consider up to LTSC 2019 to still be fully functional Vista-era NT just with a worse UI after that the changes are too much to ignore on some fundamental levels as well as app compat starts dropping off badly

i meant windows program compat.. not modern app but yeah 8.1's store is crippled last I checked

the simple fact that for now at least I can run Linux host with vmware and an acceptably performant WIndows 7 VM with samba to host where I can also share my sshfs over samba is way better than running gnome or windows 11 LOL

while my linux workflows are improving .. this VM allows me to just do stuff from muscle memory quickly .. unless its disk i/o heavy that is

So i spent the entire envining doing like every possible way to redo this clear address button.. and the selector logic got way complex so i reverted it all and just tweaked it to be on the right of the little field before the col line to the textbox it is far less jarring like. The key issue I seem to have with it is I do not expect a left corner to move when I move my mouse to do something on instinct.. but moving at that boundry between the menu

drop down and the start of the address just flows even with the slideout effect

so if I target the dropmarker symbol the target moves for the x button if I hover and it slides the right its nicer to me to use

but i don't want to just DECIDE that for everyone of course

I found the x on the left ok. Usually you hover around the middle anyway so lef or right do not matter much. Would matter in the email addy because much longer.

leaving it where it was tho just needs to constrain the image to 16x16 which is the standard size of any skin provided close icon and proper margins

suppose parity also is in play even if it kinda sucks cause outlook and thunderbird users who may happen by may be expecting the inferior microsoft design

I'll just keep it where it is and tweak the metrics slightly

frg_Away: is this sufficient? i.ibb.co/2WjFxG8/image.png

nsITobin: I've had enough of AM2-AM3 strange issues (possibly untested but supported combinations...) not to want a temperamental machine :-P

i physically cannot install an LGA processor

i don't have the dexterity

and its too pricy a risk to fuck up

i dunno what the hell I am gonna do when i go to build my next system

gigabit ethernet that causes CPU hangs on cable connection or carrier state change, plug in 8P8C - or power on connected machine - aaaand hang <3

on the other hand the motherboard could do 10base2 flawlessly via PCI card

but I lacked a 10base2 PCMCIA NIC...

greetings buc

Hi

buc ghi and happy new year

thx

frg: It would probably be a good idea to publish the source code tarball for a new release right away, without waiting for all the binaries to be ready. This will help distribution-level releases appear more or less simultaneously with the upstream.

In my experience, the issue with that is people jumping the gun in case of a last minute change.

I mean fe. 2.53.20/source/ appears first (and all others later, when ready).

buc ewong usually does this. I don't even have the credentials for the azure site where the releases are. So usually all goes up in one "load".

BTW, how many "dev people" have access to write into ttps://archive.seamonkey-project.org/releases/ ? And why?

ewong and maybe IanN.

WG9s for the nightlies

s/he/frg/

buc I probably could get access if I want but I have soo much on the platter I pass.

buc: why don't you watch the gitlab repo for new branches

already use that so the cross-reference knows where the latest branch is over time

nsITobin: I'm a bit oldschool here. :) Prefer tarballs and dislike git{h,l)ab-oriented sources in srpm.

nsITobin the source tarballs have some buildids in them so technically are only 99.9% identical. Personally this would be good enough for me.

I put up the latest windows build and checkin scripts here:

Don't bother looking for passwords or keys :)

buc: you want the true experience, spec file up an rpm to generate from hg and mq patchqueue and spit out binaries, symbols, and and a source tarball

Branch Checkout is simple: gitlab.com/frg/seamonkey-253-patche…ripts/scripts-for-official/clone.sh

nsITobin: Fedora's policies for git-oriented sources seem to be too complex (lots of spec file macros, etc.), so they are only suitable if an upatream does not provide a traditional archive at all.

s/upatream/upstream/

buc it no longer builds on CentOS 7 unless you have an updated python 3.7 and llvm 4 or higher.

frg: Nope. Successfully build 2.53.20b1 still under el7.

then the base is updated or you have at least a later python.

frg: BTW el8 systems now have rust-1.79, so packed_simd patch is needed to build with el8 system rust.

we use rustup on the builder so still use 1.73.0.

frg: or el7's python was a bit patched by RedHat, or I ignore cbindgen stuff, or something else. Just trying 2.53.20 on el7

should be overall behave better build env wise but still pieces missing.

Unfortunately, Fedora closed the possibility of el7 builds in copr.fedorainfracloud.org , so "Fedora edition" no more supports very old systems (glibc-2.17), now glibc >= 2.28 only.

I am not sure how mozilla does it but we have higher requirements than Fx now unfortunately. Suspect they build against older libs in taskcluster or we miss more build fixes. Maybe weak linking stuff missing.