github again

[AroKol78]

Serpent52 (20221216034925) + palefill1.25 links in Source File:

Error: TypeError: 'querySelectorAll' called on an object that does not implement interface Element.
Source File: https://github.com/martok/palefill
Line: 11

and

Error: TypeError: 'querySelectorAll' called on an object that does not implement interface Element.
Source File: https://github.com/martok?tab=repositories
Line: 6
Error: ReferenceError: Turbo is not defined
Source File: https://github.com/martok?tab=repositories
Line: 16

and https://github.com/

Error: RangeError: invalid value unit for option style
Source File: https://github.githubassets.com/assets/vendors-node_modules_primer_view-components_app_components_primer_primer_js-node_modules_gith-3af896-288086fb1606.js
Line: 203

[Vangelis66]

The first two reported Errors somehow are linked to the use of the new js-darktree ShadowDOM replacement polyfill...

STR:

1. Install palefill-v1.23-4-git-20221201-g677dfa9 code snapshot; that's the last with the old ShadowDOM polyfill; load: https://github.com/martok?tab=repositories result: ALL those errors are gone!
2. Install palefill-v1.24, the first where js-darktree is being used; load: https://github.com/martok?tab=repositories result:

   TypeError: 'querySelectorAll' called on an object that does not implement interface Element.

   errors start appearing 😢 ... The second error for above URI,

   Error: ReferenceError: Turbo is not defined
   Source File: https://github.com/martok?tab=repositories
   Line: 16

   isn't generated, too, until palefill-v1.24-3-git-20221203-g395dbc6 code snapshot is installed... "Turbo" = Soft Navigation and, as I've detailed here, that specific palefill code snapshot is the one that permanently breaks GitHub Soft Navigation 😞 ...

I have no clue about the last

Error: RangeError: invalid value unit for option style

error, probably a UXP-CSS-related deficiency 😞 ...

[SeaHOH]

https://bugzilla.mozilla.org/show_bug.cgi?id=1638298

A bug of the optionstyle for Intl.NumberFormat and Number.prototype.toLocaleString, style unit has being unsupported.

[Vangelis66]

The first two reported Errors somehow are linked to the use of the new js-darktree ShadowDOM replacement polyfill...

Error: TypeError: 'querySelectorAll' called on an object that does not implement interface Element.

and

Error: ReferenceError: Turbo is not defined

not present anymore when loading https://github.com/martok?tab=repositories with palefill-v1.25-4-git-20221228-g9674444 code snapshot installed... (js-darktree fixed via 3af3847)

Many thanks, have a Happy New Year! 🎅

[AroKol78]

Serpent52 (20230106125713) + palefill1.25 (+ https://github.com/martok/palefill/commit/28fceda6b56d4f614b908270bba290f5336bac3c + https://github.com/martok/palefill/commit/734c7988fe0e4daf174b15ed8c155baef2e089f2 + https://github.com/martok/palefill/commit/9674444afde74724c11242b22ec9e9a404395597)

navigation on github is terribly slow especially if you want to see the files (i navigate faster on yt and I don't have such problems - strange, to see a few lines of text requires a lot of computing power) example https://github.com/Feodor2/Mypal68/blob/main/modules/libpref/init/all.js

p.s. thank you for your contribution and work on browser development :1st_place_medal:

[Vangelis66]

example https://github.com/Feodor2/Mypal68/blob/main/modules/libpref/init/all.js

I can easily replicate on my old laptop, with a 2007-era Intel Core 2 Duo CPU T5250 @1.50GHz ("Merom") CPU... Clicking Continue once (sometimes, it appears for a second time 😞 ...) eventually makes the page load fully...

How old/powerful is your CPU?

strange, to see a few lines of text require a lot of computing power)

This isn't just text and, certainly, not a "few" lines only 😉 ... The file viewed consists of 6,069 lines of code... If you want the "text-only" version of that file, then use (s/blob/raw/g):

https://github.com/Feodor2/Mypal68/raw/main/modules/libpref/init/all.js

(loads in 1s here...).

I suppose the migration to js-darktree made things somewhat "slower" in this context... Not much can be done, probably...

[AroKol78]

How old/powerful is your CPU?

Intel Core 2 Duo 2.33GHz on the old phone with worse parameters and with the new FF- no problem

This isn't just text

for me it's just text, not quantum physics

it's just funny

[Vangelis66]

on the old phone with worse parameters and with the new FF - no problem

... Apples and oranges 😉 ...

1. "Old phone" not using Windows XP SP3 32-bit (a desktop targeting OS)
2. "New FF" (employing electrolysis by default) supports current GitHub natively, doesn't require palefill to load it (in a single process even) ...

Either martok is able to further fine-tune the palefill-included polyfills or not...

One thing is sure: MS will continue making GitHub more bloated and resource-hungry as time goes by 😠 ; they, practically, only test it on latest Edge snapshot, on Win10/11, on 2020+ H/W (which they can easily afford) ...

thank you for your contribution and work on browser development 🥇

Ditto 👍 :smile:

[Vangelis66]

Another ("tall" 😉 ) GitHub page where the "Unresponsive script" issue can be easily observed (notably on old H/W):

https://github.com/yt-dlp/yt-dlp/commit/bfd973ece3369c593b5e82a88cc16de80088a73e

[AroKol78]

Another ("tall" 😉 ) GitHub page where the "Unresponsive script" issue can be easily observed (notably on old H/W):

yt-dlp/yt-dlp@bfd973e

i tried a couple of times = Intel Core 2 Duo 2.33GHz is enough not to notice the message it doesn't change the fact that it's generally slow on github

[Vangelis66]

is enough not to notice the message

... Try also scrolling down the page (e.g. with [middle] mouse wheel) before it loads fully... FWIW, the warning can be suppressed by setting a higher (to default 20) value to pref dom.max_script_run_time ; this doesn't make GitHub load any faster, just gets the warning out of sight :wink: ...

[NintendoManiac64]

I'm slightly late to this thread but, protip from a hardware person - the Core 2 Duo is compatible with RMClock which can be used to underclock the CPU even on a laptop to slower speeds. Note that the original version lacks a signed 64bit driver, so you need to make sure to be using the variation that includes the signed 64bit driver:

• https://forums.whirlpool.net.au/archive/1287265
• https://archive.org/details/rmclock-right-mark-cpu-clock-utility-v-2.35-with-signed-64bit-driver

For Nehalem and newer Intel CPU, the same thing can basically be achieved with ThrottleStop.

For AMD CPUs, you can use CrystalCPUID for K8-based CPUs, K10Stat for K10/K10.5 CPUs, and BrazosTweaker for Bobcat CPUs, and Ryzen Master for anything Zen-based. AFAIK, there was never any OS-side underclock utility made for both Jaguar as well as the Bulldozer and its derivatives (Piledriver, Steamroller, Excavator).

[AroKol78]

Error: TypeError: navigator.clipboard is undefined
Source File: https://github.githubassets.com/assets/vendors-node_modules_stacktrace-parser_dist_stack-trace-parser_esm_js-node_modules_github_bro-327bbf-0aaeb22dd2a5.js
Line: 1

[Vangelis66]

Error: TypeError: navigator.clipboard is undefined

https://developer.mozilla.org/en-US/docs/Web/API/Navigator/clipboard

... first implemented in Fx 63, released 2018-10-23; but does it cause here any real functional breakage aside from that console entry?