02:05:56 GOOD NEWS: My patch works. 02:06:14 BAD NEWS: Adding it makes cloudflare want MORE 02:07:18 now I am getting it bitching about force-colors prefers contrast and prefers reduced motion.. and I am not entirely convinced this isn't a redhairing .. i dunno if UXP added any of these.. 02:07:45 HOWEVER if it is literally failing cause these media features it means you are being denied access OVER STYLE 02:07:53 That is ridiculous. 02:09:05 I knew css was a bad idea... ATTRIBUTES FOR LIFE! 02:09:32 02:10:58 I actually had to PORT it not just apply it manually.. the code changed SLIGHTLY and i ignored rust 02:12:02 Might be worth attempting to just.. provide the default response should actually be simpler than the prefers color scheme.. 02:12:13 it is getting late tho 08:51:21 they're using css animations too (sorry, "transitions", whatever, the thing that makes webpages feel slow and stutter), so besides webgl there's this too to fail 08:55:21 mimasama: have you had any better luck with CloudFlare? 12:58:43 Winders is now all setup to build the seamonkey more cleanly than yesterday.. I am gonna redo the UXP prefers-color-scheme patch and attempt to incorperate tests from the stylo version.. we at least get that out of this.. 13:00:46 The old Windows Update dialog was DESIGNED to operate sequantually.. its acting pretty off since in windows 10 by this point they made em all download at once 13:47:57 nsITobin thanks. don't bother with tests. You would need to build Firefox for them but because of the python changes there is breakage there currently. wip fixes some of it already but needs more. 13:47:59 I would make it a TOP- patch. If we get stylo up later it could be replaced and if we decide to scrap stylo we can integrate it properly later. 13:48:37 http://files.thereisonlyxul.org/build-tools/mzlabld-env/scripts/start-shell-vs2022.bat 13:48:38 btw 13:49:16 it has a snag because it ises vswhere like the build system it will find the latest version 13:53:53 well I am falling back to windows for primary seamonkey development because as it stands doing dev on linux is painful at the moment. My environment is a bit too new for it.. an issue I intend to help improve of course but have to have a reliable way to build it to make it reliable everywhere 13:56:08 option 1 of course skipps the older vcvars detection and calling and is eq to just start-shell.bat so you can use both as you see fit .. if you want it to be for 2019 just change the year in the script. 13:58:10 frg_Away: is stylo hooked up? 13:58:34 like at all? 13:59:00 nsITobin disabled per default. If you enable it breaks. Needs more stuff ported and with rust it is a rat race. 14:01:48 it would be IDEAL tho gross if we could hook stylo up and leech off it till the actual style system can get cpp patches 14:01:52 nsITobin is it needed for normal building order bootstrap only? Currently I specify ac_add_options --with-visual-studio-version=20xx in the mozcinfig and it picks up the toolchain. VS2017 VS20919 and VS2022 currently supported. 14:02:24 that's right... you have that capability 14:03:45 I couldn't get sm to pick up the .mozillabuild artifact version of vs2019 14:03:53 nsI yeah we added everything essential up to 60 and more but ripped out some gerated stuff and this needs more later fixes. Hacked cssparser to be buildible under later rust. So cssparser cbindgen and other stuff need to be updated first. 14:04:41 nsITobin it calls vswhere internally and the artifact is probably not registered. So a no go with the artifacts. 14:04:49 my dev environment is recreated anytime something changes so I hit snags you may not if you been using something established 14:05:18 which is good cause can squash bugs! 14:05:19 http://xr.thereisonlyxul.org/seamonkey-2.53/source/build/moz.configure/toolchain.configure#562 14:06:13 the batch file uses pretty much the exact same logic but where is it using the state directory 14:06:34 nsITobin well I like it rather static :) Have 3 build vms with different configs and WG9s does clang daily so usually Windows bugs are covered fast. Linux on the other hand... 14:06:59 need a coffee bbl 14:08:53 oh yeah Windows seems to run well 14:09:06 windows version 14:14:11 https://xref.palemoon.org/goanna-central/source/platform/build/moz.configure/toolchain.configure#400 14:14:20 a200 didn't do a very good job with this 14:15:02 and yucksp didn't get that configure flag 14:26:45 back 14:30:37 I took almost the whole buildchain patches till 65 at this time I think. Just adapted for VS2017 and up which mozilla never did properly 14:31:16 still good time where I didn't need to do firefighter every day and could concentrate one on task. 14:31:34 I think I basically merged your logic from it with a200's for aura .. for the five minutes that existed over several months... 14:32:14 I dunno HOW I got switched to DBZ Nemekian Time but I need to go faster! 14:33:25 decided to index waterfox.. still debating if I should tell him I am so he can benefit 14:34:41 greetings therube 14:35:03 therube: fyi logbot.thereisonlyxul.org/seamonkey 14:35:11 exists 14:37:03 oh my god.. I figured it out.. Waterfox is literally setup like a built in xul extension 14:37:09 LOL 14:37:36 actually that isn't a bad idea but it sure relies on some hacky hacks 14:40:04 if that stuff were still in place aside from rebranding firefox .. everything waterfox is is the eq of one of the more complex xul extensions. 14:41:07 I kinda approve now that I get it 15:03:26 one of my secondary monitors blanks out when building the code 15:07:45 https://ibb.co/9hqnmVq 15:07:57 bugs me 15:09:18 bbl 15:09:22 k 15:10:49 I see a few style issues there but the aren't difficult to fix on windows at least 15:13:47 cloning heptapod 15:13:54 this is gonna take a good minute 15:14:26 be back as well 15:50:08 I wish dom still used xpconnect instead of jsapi 15:51:57 nsITobin Bug 1909483 15:56:28 fun! 15:58:17 mxr needs a real big cleanup and push to move most functions to PHP 15:59:00 my version won't run on old apache/cgi anymore 16:00:03 good thing too last time it was four requests to certain files would bring the server down 16:01:34 sqlite.c is completely excluded from search and ident because its single file nature is just too much to handle past like a handful of requests.. everything else save perhaps old-configure does pretty well considering its age 16:02:07 however I did tell it to treat sys.mjs and mjs files as javascript 16:03:11 rust support I don't see happening very easily but I should be able to futz in some syntax highlighting 16:05:14 getting some heavy rain and some lightning.. hope i won't have to shut down 16:05:22 or the internet cuts out 16:06:24 I am getting warnings about watchman is that normal? 16:06:31 when applying 16:12:54 nsITobin I took watchman.exe out the hard way by renaming it. Sucks up cpu for nothing really needed. 16:13:56 so how do i make this heptapod tree build? 16:15:15 nsITobin which one? 16:15:46 253 16:16:03 https://dpaste.org/6Mpgj/raw 16:16:53 stock mzlabld 3.4 16:19:18 Min works but updated. 16:19:37 Try export MACH_USE_SYSTEM_PYTHON=1 in the mozconfig 16:20:29 seems happy with that 16:21:05 yeah these are all things I should include in updated build instructions 16:22:08 nsITobin 2.53 is a bit wobbly. wip is further with build fixes. Might need to merge them first if I can make wip to fully work in time 16:23:21 this one? https://gitlab.com/frg/seamonkey-wip/-/tree/master/frg/work-js?ref_type=heads 16:24:25 I will let this build complete so I know it is good and do it again with this repo 16:32:33 nsITobin yes 16:33:38 nsITobin branch next-level-stuff https://gitlab.com/frg/seamonkey-wip/-/tree/next-level-stuff?ref_type=heads 16:34:17 you need to rename the series-text files to series or just ignore the error at the end if you do qpush -all 16:34:49 I put a non existig patch in to shield the not yet done stuff 16:35:13 l10n is from 2.53 16:38:03 well what I built identifies its self as 2.53.20 16:38:10 is that correct? 16:38:17 for the 253 patches 16:38:20 repo 16:38:38 nsITobin yes 16:38:46 oh beta 1 pre as well 16:38:58 yes again 16:39:21 dropped all the stuff for the ua 16:40:57 $ cat comm/suite/config/version.txt 16:40:58 2.53.20 16:40:58 $ cat comm/suite/config/version_display.txt 16:40:58 2.53.20 beta 1 pre 16:41:06 $ cat comm/suite/config/version_package.txt 16:41:06 2.53.20b1pre 16:41:51 you not using toolkit abpre+ stuff as to not trigger the build system into building nightly or beta code? 16:42:18 in version.txt 16:42:38 ? 16:46:38 correct 16:47:18 not even sure it builds with enable-nightly. 16:48:09 not up for depthcharge builds? 16:48:27 yes 16:48:37 i always thought seamonkey should have called em that in the minefield era 16:48:39 want stabilty 16:49:20 If we find something we want we drive it till nighty pref for it is gone. 16:50:11 took years to resolve all the codepaths in UXP but mostly it took years to get anyone to do it lol 16:51:17 there is also enable-debug and enable-fuzzing and a 100 others 16:51:22 granted esr52's divergence with nightly early alpha beta/release was a lot smaller than it quickly became 16:51:55 enable-debug is the only one I should take a look when I find some time. Probably broken. 16:52:42 again familar debug builds were busted for years 16:53:12 They are basically unusable even if working. 16:54:36 scratch that debug builds WERE working initally and they got busted for years from not being maintained then fixed at my urging.. I had to use debug output to port Pale Moon's appcore enough to get a main window so it could be fixed the rest of the way 16:55:26 as for some reason console output isn't enablable even basic console output isn't enablable on windows in non-debug builds 16:55:50 and the js console was removed and devtools couldn't work cause not firefox 16:56:13 I actually don't wanna think about that.. it was annoying lol 16:57:56 nsITobin well we have both consoles :) 16:58:20 of course.. who do you think I stole the updated code from? 16:59:01 devtools is better but if I want to see whats going on during start I need the other. 16:59:34 SeaMonkey has enough consoles that I should probably finally figure out which one is better for which task 17:00:19 in 52 devtools ran like crap for a long time till certain fixes in the js engine improved it quite a bit.. https://repo.palemoon.org/MoonchildProductions/UXP/issues/338 17:00:33 sorry they run like crap NOW they USED to run like shit 17:00:34 lol 17:00:42 under UNIX there can be stuff printed to the standard streams in the associated terminal device, under windows I often enable the same, although the windows' terminal emulator feels a bit lacking, then there is the error console, and was it two or three more? 17:01:12 a js console (devtools?) and a browser console in devtools style too? 17:01:14 njsg: for causal errors and chrome errors .. toolkit js console.. for web unless you have that hack enabled.. devtools 17:01:34 is my opinion 17:01:38 fight me lol 17:02:06 oh hell i forgot about the BROWSER console 17:02:07 trickiest thing for me was to remember that, to get the command prompt in the browser js console, that needs a setting (debugging?) enabled 17:02:14 which is chrome level devtools error console 17:02:16 so we have THREE 17:02:20 ALL THE CONSOLES! 17:02:38 at one point I was suggesting that to run a bit of code with access to XUL windows, and wasn't aware it'd normally not allow that 17:03:00 * njsg plugs in the VT and directs output there - does it count as a fourth? 17:03:20 console console is FOUR 17:03:25 but not on windows unless debug 17:04:17 I don't think a debug build is necessary, just one or two prefs to enable that output? 17:04:22 SeaMonkey the All in one Console Suite 17:04:46 njsg: well if there is on windows i never found it so please share 17:05:44 Elect me as President of the United States.. every component will be matched with a slightly different console 17:06:39 could be mozilla added something as well .. heptapod is timing out on me 17:07:15 nsITobin: I think -console parameter in seamonkey.exe then some setting which I don't remember right away 17:07:59 i think you won't get any stdout on windows of course right now heptapod is http request error incomplete response 17:08:20 I could swear there's an option for that but I can't find one modified in this profile I have open :-\ (maybe it's necessary only on windows?) 17:08:25 nsITobin wich hg? stock 3.4 might be too old 17:08:35 it checked out earliuer 17:08:43 but I'll use systemhg 17:09:01 uhhh winkey+r cmd 17:09:03 do 17:09:05 python -m pip install --upgrade pip 17:09:06 pip install -U appdirs 17:09:08 pip install -U compare-locales 17:09:09 pip install -U fluent 17:09:11 pip install -U mercurial 17:09:12 pip install -U packaging 17:09:14 pip install -U pyparsing 17:09:16 pip install -U setuptools 17:09:17 pip install -U six 17:09:18 pip install -U virtualenv 17:11:53 it errored out at mercurial 17:11:55 frg_Away: 17:12:43 I overcame it but maybe because of bootstrap esr115 17:13:07 i already ran bootstrap for esr115 and downgraded rust 17:13:28 ERROR: Command errored out with exit status 1: 'd:\binoc\workstation\applications\Environment\smbuild\python\python.exe' 'd:\binoc\workstation\applications\Environment\smbuild\python\lib\site-packages\pip\_vendor\pep517\_in_process.py' get_requires_for_build_wheel 'c:\users\mattat~1\appdata\local\temp\tmpoouzzd' Check the logs for full command output. 17:18:28 I am using system installed hg to checkout 17:18:49 should perform some surgery on mozillabuild me things 17:31:06 abort: HTTP request error (incomplete response) 17:31:06 (this may be an intermittent network failure; if the error persists, consider contacting the network or server operator) 17:32:02 nsITobin yeah can reproduce in anothervm.there at 5.92 17:32:22 oh okay i was starting some network diagnostics lol 17:32:32 cause seems like it ain't the local workstation 17:34:50 are these patches successive as in if I had to could I do it with git? 17:41:58 i wonder where their datacenter is 17:51:16 oh good it cloned down 17:52:24 could be the storms if they are dc'd in texas or louisiana 17:54:59 nsITobin successive copy to .hg/patches and just run hg qpush --all 18:09:04 unable to read 253-release-top.patch 18:09:04 now at: 1470307-1-PARTIAL-BACKOUT-25320.patch 18:09:07 normal right? 18:09:18 for comm 18:09:41 nsITobin yes that is my marker Stuff beyond is for later or never 18:29:37 i need more esr trees.. thought i could put it off for a while but nop 18:30:59 I only need TortoiseHg and can look up the history of directories and files easy with it. 18:31:23 grab it. It is a lifesaver on Windows and Linux 18:37:13 i have it 18:37:44 i just need to learn it better 18:39:40 still patching the platform 18:39:48 takes a good minute eh? 18:39:49 have set it up with console and repos: https://ibb.co/bmf2fCH 18:40:43 yeah 18:40:52 nsITobin yes when this is ready we need to update gitlab and heptapod masters. Currently about 11000 patches in mozilla and this is a bit much. 18:41:04 hg is why I originally started using sourcetree 18:41:53 but the lesson there is sourcetree did not really teach me how to use git or hg 18:42:10 of course workbench is a bit lower level than that old junk 18:42:16 from atlassian 18:56:49 wip build kicked off 19:15:13 sorry for the abrupt exit i checked my cpu temp and had to shut down and clean the dust out of my cpu cooler 19:18:51 and my environments broken 19:20:46 frg_Away: remind me once I have it figured out and reproducable to recompile the vmware modules and create a virtual machine 19:21:05 nsITobin if this is abotu the congoing build. I usually rename python3 to python3something in mozilla-buld. Usually kills the build gracefully. 19:21:48 well it WAS building but I had to shut down when I restarted it was failing I think I just need to refresh my mozillabuild 19:23:47 there it goes 19:23:57 yeah python got corrupt or some shit 19:24:10 I think I am gonna have to NOT access this drive from linux anymore 19:25:57 kicked off again let's hope my cpu stays under 80 instead of damn near 90 19:26:00 C 19:27:22 under "normal" conditions my processor should hit the power limit before hitting the thermal limit which seems to be about 82-83 when it shot up to 89 well dust. 19:27:43 frg_Away and nsITobin: have you ever tried a Windows Docker Container with Visual Studio 2019 or 2022 and all necessary tools to build SeaMonkey ? this docker file can than easily be shared by developers to have a working build environment without risking the local computer for daily use ? 19:28:11 guest_: does one currently exist? 19:28:22 you could than easily play with different Rust version without risking to loose a working build environment 19:28:22 guest_ no 19:29:11 see we have a bootstrap issue here including mozbootstrap lol we need to be able to one hundred percent need to be able to create a build environment when needed to create a build environment that could be shared.. HOWEVER.. what about windows licenses? 19:29:40 images with VS 2019 or VS 2022 should be available and they need to be updated to include all other needed tools for SeaMonkey build 19:29:54 guest_: visual studio isn;t the only question 19:30:07 some of these bits and bobs are on the fly 19:30:50 I currently use virtualbox vms. Zero experience with docker and I suspect will be the same when I cross the big bridge. I unfortunately don't have the time to go this route and set it up. 19:31:16 throw rust deps in it and we'd spend way too much time updating images than updating the code.. it's a great idea if the issues can be worked out but right now.. build instructions that anyone cna follow and it work for em on their respective os would be far more useful and attainable. 19:32:04 that's the point - there's no up-to-date build instructions for Windows. 19:32:21 I will be updating some soon as I figure out everything ;) 19:33:11 everything which had been configured in the VM should be able to replicated in dockerfile 19:33:13 any one of us can bruteforce a working env.. kinda what I am doing right now but that doesn't help someone like mrnhmath 19:35:29 guest_: this is not a new concept .. mozilla used to have some images at one time.. and I made a centos 7 vm and shared it so we all had a common build setup.. the issue right now is everything is in major flux and getting EITHER easy old style dep find and get OR new style half-automatic .. we have to nail down what needs what and write it down before any such image could even be thought about ;) 19:35:43 frg_Away: what Windows version is your VM running ? 19:36:45 Server 2012 R2 for release 2016 for most of the work 2019 for 2.53 only and 2022 for future. 19:37:22 I build on a lightly NTLite'd Windows 10 LTSC 2019 19:38:29 hope they come back they have a good idea surely worth putting on the want list 19:38:37 our next build server uses 2019 19:40:58 is this switched over to components.conf or are they there just for patches? 19:42:03 just for the patches. Don't think it is used. 19:43:09 wait was this compiled by vs or clang? 19:43:36 about:buildconfig shows only clang 19:44:55 nsITobin if it finds a working clang it picks it. Have not fixed it. Use export CC=cl in the mozconfig. 19:45:20 well it built 19:45:25 and .. runs with clang 19:45:53 i'd really like to know what we need visual studio for if clang can compile 19:46:21 what is it used for in modern mozilla? 19:47:03 nsITobin support stuff includes and other things. 19:47:16 core config https://paste.debian.net/1324145/ 19:48:16 build config which includes it: https://paste.debian.net/1324146/ 19:49:33 simple shell script: https://paste.debian.net/1324147/ 19:53:42 and now sleep. nn 19:53:49 rest welll 19:55:16 here is the one I am using https://paste.debian.net/1324148/ 20:03:30 solid 75C 20:04:05 njsg: doesn't look like I will be burning my computer out building seamonkey .. today anyway :P 20:59:46 nsITobin: it's not khtml-chromium-qtwebkit, for sure 21:00:20 or, how I call that one, "stress test that can trigger thermal shutdown" 21:35:05 qtwebkit was a respectable project .. qtWEBENGINE is a big fat lie 21:35:16 as just wrapped chrome 21:35:32 where qtwebkit was an implimentation in webkit if i am not mistaken 21:35:41 of webkit in qt 21:36:41 WebKit is an open source web browser engine. WebKit's HTML and JavaScript code began as a branch of the KHTML and KJS libraries from KDE. As part of KDE framework KHTML was based on Qt but during their porting efforts Apple's engineers made WebKit toolkit independent. QtWebKit is a project aiming at porting this fabulous engine back to Qt. 21:37:07 and its dead 21:38:15 njsg: can you give me a broad overview of how to deal with l10n cause if I add the ui bits of prefers-color-scheme the strings will need updated in not-en-US 21:43:26 someone tried to revive it but of course couldn't keep up 21:44:12 well mrnhmath there you go.. port seamonkey to qtwebkit 21:44:16 :P 21:45:29 port seamonkey to ladybug 21:45:32 whatever that project name is 21:53:00 nsITobin: non-en-US locales are in the l10n repository, so you could make patches to update that one. there might have been ways to deal with this via pontoon, not sure what's the best approach now 21:54:38 nsITobin: actually, isn't there a queue for l10n too? 21:55:29 right i worked that part out.. question IS do I just add the english strings or is there a process to get em translated i have to wait on .. or?? 21:56:31 ah ok, not sure about that 22:01:21 I personally would want prefers-color-scheme in the next release cause it is a low risk patch that DOES have beyond clownflare webcompat impacts .. why do you think it was added to UXP when there are other nifter things like fuckin background bluring 22:01:40 but it has l10n impacts IF the GUI bit is added 22:02:23 idealy it would be nice to expand it so it CAN follow system scheme as well AND even make it more of a site preference so you can have light or dark on a persite basis 22:02:35 for instance i prefer some stuff dark and other stuff light 22:03:06 but wait long enough and a FORM of what I just said will be in eventually ;) 22:03:09 just not as good 22:03:52 is there a prefers-color-scheme: black-no-I-really-mean-000000-not-some-residual-brightness? 22:04:05 but hell if stylo isn't an immidiate consern.. then we can certainly mine UXP for style patches largely applicable to seamonkey's nsStyle 22:04:29 dark gray actually has a visible impact in the background here, so being able to choose 000000 would improve matters *so* much in some sites 22:04:50 njsg: no unfortently .. despite amoled being more commonplace getting that sliky smooth black just ain't in the current design trends 22:04:58 but I guess it's one of these things, like how I get more readable text the instant I disable webfonts :-P 22:05:17 nsITobin: ah I see [turns on VT and fires up elinks] 22:05:22 I'd exterminate for a huge ass amoled screen 22:06:17 dark mode? you mean monolith mode 22:06:22 njsg: That 22:06:28 that is what we should call it 22:06:32 light dark monolith 22:06:34 ;) 22:07:55 I'd go with monochromatic or monochrome except even that one isn't much of a good name, usually it has at least two colors (foreground and background) *and* you can get two more (standout and bold), so not exactly *mono*chromatic 22:10:19 (if anyone's wondering, foreground and background are usually swapped in reverse video, so there you'd get these two colors (or one color and lack of color, okay, that one might be suitable for "monochromatic"), now bold can be a different brigtness (or, in a terminal emulator that does colors, even a completely different color); standout mode is reverse video but with that color as background) 22:10:28 all of that much more readable than gray on white 22:10:46 monodramatic? 22:11:01 no, we always go with THX 22:11:07 wasn't it black on grey already? 22:11:17 originally 22:11:20 in the old days 22:11:26 that was netscape's doing wasn't it? 22:11:48 that was probably relatively common, because pure white as a background wasn't much of a good choice 22:11:54 NCSA Mosaic did that for sure 22:12:59 Windows in general used to do it, at least with Windows 4 and I'm guessing NT at least 4? or 5 thru 6.1? Not really sure about newer ones. 22:13:11 #d4d0c8 in NT 5.0