Jump to content

My Browser Builds (Part 3)


Recommended Posts

Yay, more incompatibilities (Stack Exchange sites, Tumblr...). I tried transpiling that MSFN service worker, it still didn't work.

I've held on for a very long time, but even from my POV, this ship has sunk unless the developers behind these browsers step it up.

Until then, as Agent Smith would say:

spacer.png

Late edit: Seems inline (in HTML) incompatible JS might be only on some Stack Exchange's sites, at least on Stack Overflow for now.

Edited by UCyborg
Link to comment
Share on other sites


23 hours ago, UCyborg said:

Yay, more incompatibilities (Stack Exchange sites, Tumblr...). I tried transpiling that MSFN service worker, it still didn't work.

I've held on for a very long time, but even from my POV, this ship has sunk unless the developers behind these browsers step it up.

Until then, as Agent Smith would say:

 

Late edit: Seems inline (in HTML) incompatible JS might be only on some Stack Exchange's sites, at least on Stack Overflow for now.

just noticed UXP contributor is working on opt-chaining syntax:
https://repo.palemoon.org/MoonchildProductions/UXP/issues/1658#issuecomment-29898

Link to comment
Share on other sites

6 hours ago, Elkern 4926 said:

You should see if moonchild will add xp support back to Palemoon now that one guy is gone.

Is that guy still screaming at those meddling kids?

Link to comment
Share on other sites

On 5/2/2022 at 6:57 PM, UCyborg said:

Yay, more incompatibilities (Stack Exchange sites, Tumblr...).
I've held on for a very long time but, even from my POV, this ship has sunk

Add
https://www.cloudflare.com
to the long list of sites that end up loading a blank page (latest St52 32-bit); CF homepage produces several TypeErrors in the Web Console... :angry:

But seriously, today I jumped into a real deal breaker for UXP-based browsers: GitLab; it is no longer possible to sign into an existing GitLab account on a UXP-based browser (plus some others, read below) :realmad: :realmad::realmad: ...

To begin with, to even successfully load GL pages in a UXP browser you're in need of the gh-wc-polyfill extension (manually/forcibly) installed (GL break things ever so often :angry: , make sure you're on the latest beta/stable release - currently v1.2.19). As long as you're not logged-in, things appear OK...

STR:
1. Load, as a guest, e.g.,
https://gitlab.com/cleanflash/installer/-/releases
(happens to be one of my bookmarks ;) )

FiuAvWQ.jpg

2. Click on the "Sign in / Register" button (far right, top header); GL will load
https://gitlab.com/users/sign_in/

3. This is the crucial step where things broke :realmad: ; to verify you're not an automated bot trying to sign-in, GL are serving your client (browser) with a CloudFlare-based "challenge" that it has to pass successfully in order for the sign-in page to display: 

fV9Ep8K.jpg

During the last week(s), that "challenge" has been "upped" and is no longer compatible with UXP browsers :realmad: ; the browser goes into an infinite loop of those "5 seconds" redirection cycles, "frying" your CPU in the process :angry: , and never affords the GL sign-in page below:

kogs5bx.jpg

FxESR 52.9.0 and (latest) Serpent 55 behave similarly (to St52) on my Vista SP2 32-bit machine; but I also gained access to a Win7 SP1 64-bit laptop and can report that Fx 56/ FxESR 60.9.0 / FxESR 68.12.0 are also unable to generate the GL sign-in page there; however, and that's really no surprise, the latest Fx 100.0 (32-bit) passes the GL challenge with flying colors and displays the GL sign-in page in ca. 2 sec ... :dubbio:

The "challenge" is probably "feature"-based, rather than "UA"-based, because I tried to spoof "working" Fx/Chrome versions in Serpent 52, sadly with no success :( ... 

Back to the Vista machine, I also tried 360EEv11/12; the former (v11, Chromium 69 based), does require 1 to 3 redirection cycles (ca. 5-15sec, YMMV), but eventually does pass the GL challenge and the sign-in page is shown :P; the latter (v12, Chromium 78 based) passes the "challenge" more effortlessly and affords the GL log-in page very quickly (the screenshot above is with that browser) :thumbup ; I then used my GitHub account credentials to successfully access my GL account in 360EEv12 (I did not bother checking v13, I'm certain it will, too, work OK :P ). 

Just as a POC, I then exported (via an extension) ALL GitLab cookies from 360EEv12 to a "cookies.txt" file (Netscape format) and subsequently imported them (via another "legacy" extension) into latest Serpent 52; and that is the only way (obviously impractical :angry: ) I could browse GL being logged-in with St52:

WWnwyIa.jpg

If anyone knows of a way to make GL log-in work natively in latest St52, do come forth! :P

 

Edited by VistaLover
Link to comment
Share on other sites

17 minutes ago, VistaLover said:

The "challenge" is probably "feature"-based, rather than "UA"-based, because I tried to spoof "working" Fx/Chrome versions in Serpent 52, sadly with no success :( ... 

This has been the norm as far as the "modern web".  Spoofing UA is basically a waste of time nowadays.

Link to comment
Share on other sites

On 5/3/2022 at 6:09 PM, roytam1 said:

just noticed UXP contributor is working on opt-chaining syntax:
https://repo.palemoon.org/MoonchildProductions/UXP/issues/1658#issuecomment-29898

This has now been merged into the master branch of upstream-UXP:
https://repo.palemoon.org/MoonchildProductions/UXP/commit/92b3def6a88f9ea15c9e51e22a57b667edea29c0

Hopefully, it will also arrive in this weekend's roytam1 builds :P (fingers-crossed it doesn't break other aspects of the browser :dubbio:) ...

Link to comment
Share on other sites

I have a question about the "keyword" feature of bookmarks. Take the default google quicksearch as an example:

Set a bookmark as "https://www.google.com/search?q=%s" with the keyword "g", then simply enter "g abc" in location will visit "https://www.google.com/search?q=abc". A feature I used much, at site like google, webarchive, etc.

The problem is, sometime I need that %s variable contain the "/" character, which is always replaced by "%2F" in Serpent 52. I can recall it used to work (maybe in Fx 52), but I can't remember whether the "/" character remains as is,  or the site decodes "%2F" back to "/".  Can that behavior be changed in browser?

Link to comment
Share on other sites

15 hours ago, VistaLover said:

This has now been merged into the master branch of upstream-UXP:
https://repo.palemoon.org/MoonchildProductions/UXP/commit/92b3def6a88f9ea15c9e51e22a57b667edea29c0

Hopefully, it will also arrive in this weekend's roytam1 builds :P (fingers-crossed it doesn't break other aspects of the browser :dubbio:) ...

sure, they're in my tree now. with some preliminary testing (i.e. stackoverflow) it seems to be working.

nullish coalescing is still not in any UXP tree so have to wait.

Link to comment
Share on other sites

Hello all!

Do you guys feel there is benefit to leave the pref:

plugins.rewrite_youtube_embeds at its default "true"

?

Yes, I understand I'm likely one of the few here still trying to watch videos in the browser without Polly Fill ... just figured I'd check to see what others think whether or not it could slow things.

Thanks in advance :)

Edited by XPerceniol
Link to comment
Share on other sites

37 minutes ago, XPerceniol said:

plugins.rewrite_youtube_embeds at its default "true"

Dearest @XPerceniol :); my own stance is for people to stop meddling with "about:config" advanced settings, especially when not sure what they're about; there's a good reason why "general.warnOnAboutConfig" is by default set at "true" :P ; but this is just me getting grumpier as I grow older... :P

The referenced pref is an old remnant from an era when yt were still using both Flash and HTML5 embedded players; more at this link ;) ; for many years now, yt are exclusively using their HTML5 player, so the setting of that pref is currently moot: you'll always be served the HTML5 version of the embedded yt video... 

I briefly toggled the pref to false, just to humour you, no yt breakage encountered here...
As for performance slowdowns, am afraid this is something you'll have to check on your own particular setup (H/W+S/W) ...

Best greetings :)

EDIT: @UCyborg beat me to it by mere seconds! :P

Edited by VistaLover
Link to comment
Share on other sites

3 hours ago, VistaLover said:

Dearest @XPerceniol :); my own stance is for people to stop meddling with "about:config" advanced settings, especially when not sure what they're about; there's a good reason why "general.warnOnAboutConfig" is by default set at "true" :P ; but this is just me getting grumpier as I grow older... :P

@UCyborg beat me to it by mere seconds! :P

Oh, some of the grumpiest older people are the very people I would rather seek advice from ... so no worries and I'm afraid (and hate to admit) not only is the only warranty I void, I've removed the tags from all pillows and mattresses ( they're coming for me for sure now ). 

Thank you both @VistaLover and @UCyborg for helping a fellow grumpy blobfish with his endless and (sometimes/often times) silly questions.

:D

I will try to leave things alone, but I'm always like, *if I push this button, what does that do*. Ya know, sort like like kid with the cookie jar :)

Now when I press the rewind button on my universal remote control the garage door opens and closes - seriously looking forward to future Serpent development by Roytam.

Edited by XPerceniol
Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...