Google-Chrome 'portable' 64-bit browser - (with updater!)

[rockedge]

I am having the problem with the updater I think. It seems what happened to @mikewalsh started happening to me on this setup. The update prompt GUI keeps appearing and will run the update but the /chrome directory is deleted when the update process is finished. I watched in a terminal the launch script run and it appears the updater does the wget successfully and the updater message reports a successful update. But the entire /chrome directory is gone.

So now when I run LAUNCH it says file or directory not found and stops. So to sum it up it seems the updater script mechanism is behind the issue

When I copy the /chrome from the tarball back in the LAUNCH from mikewalsh and the one with the modification both start up Chrome portable

[mikewalsh]

@rockedge :-

I shall have to have a look at the updater script over the Easter weekend, Erik; see if I can figure out what's going on.

Leave it with me.

Mike.

[spiritwild]

I had the same issue. Had to look through the script and just download it manually. Copied the directory manually.
I just figured it wasnt compatable with pup devuan.

[mikewalsh]

FWIW:-

I've just successfully updated to 100.0.4896.127, following reports of the critical CVE reported earlier today. No issues at all.

I'm hitting the sack. Been a looong day.....

Mike.

[mikeslr]

Thanks, Mike, for the update.
I also had had the reported problem of the disappearing chrome folder on update. Thought it might have been a re-occurrence of something like the need for 'ar'. But this update to version 100.0.4896.127 went as smooth as silk. Posting from it now.
Something strange about the intermediate version just before 100.0.4896.127?

[spiritwild]

I noticed that as well with this update. mine updated without issue this time. must have been a glich in the matrix.

[mikewalsh]

I noticed that as well with this update. mine updated without issue this time. must have been a glich in the matrix.

Mm. I have to agree. It's only the last couple of builds that have misbehaved themselves, TBH. Hopefully, things have "settled down" once again.

We shall see what there is to see....

Mike.

[Flash]

@mikewalsh the link in your post here is 404 dead.

[mikewalsh]

@mikewalsh the link in your post here is 404 dead.

Sorry, Flash. I deleted that one quite some time ago.

Although it's possible to still RUN Chrome as root, you have to use the '--test-type' ''switch".....and as others found out, certain websites won't let you sign-in then. No Google apps will work, for instance.

I decided it was more trouble than it was worth, and gave up on it. I'll modify that post to make this obvious.

Mike.

[user1234]

I had posted 1 issue in fossapup64 forum, with title 'removing google-chrome also removed...'. I actually now know that the cause of the problem, i.e. ar, libbfd.so and libbfd-2.34-system.so getting removed every time after I run update on chrome-portable.

I tried to dive into your UpdateChrome code and found exactly what I had expected. You first symlink the ar and other libbfd*.so files without checking if they were previously loaded or not (i.e. devx.sfs was already loaded or not), and then remove them without checking as well-

Code: Select all

#
# Determine Puppy release & copy relevant 'ar' files...
#
if grep -q 'Quirky April64' /etc/DISTRO_SPECS
then
	ln -s $HERE/DATA/Ar/Tahr/ar /usr/bin/ar
	ln -s $HERE/DATA/Ar/Tahr/libbfd-2.24-system.so /usr/lib/libbfd-2.24-system.so
fi
#
if grep -q 'puppy_xenialpup64' /etc/os-release
then
	ln -s $HERE/DATA/Ar/Xenial/ar /usr/bin/ar
	ln -s $HERE/DATA/Ar/Xenial/libbfd.so /usr/lib/x86_64-linux-gnu/libbfd.so
	ln -s $HERE/DATA/Ar/Xenial/libbfd-2.26.1-system.so /usr/lib/x86_64-linux-gnu/libbfd-2.26.1-system.so
fi
#
if grep -q 'puppy_bionicpup64' /etc/os-release
then
	ln -s $HERE/DATA/Ar/Bionic/ar /usr/bin/ar
	ln -s $HERE/DATA/Ar/Bionic/libbfd.so /usr/lib/x86_64-linux-gnu/libbfd.so
	ln -s $HERE/DATA/Ar/Bionic/libbfd-2.30-system.so /usr/lib/x86_64-linux-gnu/libbfd-2.30-system.so
fi
#
if grep -q 'puppy_fossapup64' /etc/os-release
then
	ln -s $HERE/DATA/Ar/Fossa/ar /usr/bin/ar
	ln -s $HERE/DATA/Ar/Fossa/libbfd.so /usr/lib/x86_64-linux-gnu/libbfd.so
	ln -s $HERE/DATA/Ar/Fossa/libbfd-2.34-system.so /usr/lib/x86_64-linux-gnu/libbfd-2.34-system.so
fi

Code: Select all

#
#Remove 'ar' files
#
if grep -q 'Quirky April64' /etc/DISTRO_SPECS
then
	rm /usr/bin/ar
	rm /usr/lib/libbfd-2.24-system.so
fi
#
if grep -q 'puppy_xenialpup64' /etc/os-release
then
	rm /usr/bin/ar
	rm /usr/lib/x86_64-linux-gnu/libbfd.so
	rm /usr/lib/x86_64-linux-gnu/libbfd-2.26.1-system.so
fi
#
if grep -q 'puppy_bionicpup64' /etc/os-release
then
	rm /usr/bin/ar
	rm /usr/lib/x86_64-linux-gnu/libbfd.so
	rm /usr/lib/x86_64-linux-gnu/libbfd-2.30-system.so
fi
#
if grep -q 'puppy_fossapup64' /etc/os-release
then
	rm /usr/bin/ar
	rm /usr/lib/x86_64-linux-gnu/libbfd.so
	rm /usr/lib/x86_64-linux-gnu/libbfd-2.34-system.so
fi

This causes the ar and other files to get deleted, even if they were located there already. This can surely be called a 'bug', rather a 'main' one.

I only use your chrome and zoom portable, so only know about their Update files, and they do have got the 'bug'. Please remove this bug from all of your protables.

[mikewalsh]

@user1234 :-

A-HA. Good catch.

Yes, I missed that one when I cobbled the update script together. (Bad Mike..!)

Leave it with me. It shouldn't take too much sorting out; I've equipped 3 items with the updater in total.....Chrome-portable, Iron-portable, and Zoom-portable. It'll probably be after the weekend, but might be sooner.

The same 'fix' will be needed in every case, so will only need figuring out once.

Cheers!

Mike.

[spiritwild]

how did chrome become so portable? What changed?
Years ago, you had to have a special wrapper script, sandbox and 14 beers to make that thing work.
I avoided it for the longest time.

[mikewalsh]

@spiritwild :-

how did chrome become so portable? What changed?
Years ago, you had to have a special wrapper script, sandbox and 14 beers to make that thing work.
I avoided it for the longest time.

^^^ Lolololol !!!

I got round it by creating a dedicated, mini-'spot' directory inside the portable directory. That fixed the 'bitching' about the sandbox. Still needs the "special" wrapper script, though.

As for the 14 beers....

Mike.

[mikewalsh]

@user1234 :-

Right. Now then; I believe the updater script should behave itself now.

Early in the script, it now checks to see if 'ar' or 'libbfd' are already installed. If they are, it skips this section and proceeds with the updating.

Toward the end, when looking to remove those items, it now ONLY looks for sym-links. If they've been installed separately, or as part of a .pet, or by loading the DevX SFS, they will show up as regular files. Only if they've been 'linked-in' by the script will they show as sym-links.....so the script will only remove them if this is the case.

In modified form, the script is working A-OK for me. I've attached the 'Update' scripts for both Chrome & Zoom to this post; if you want to try them out, just swap them with the old versions, then see what happens on the next update.....OK?

Let us know how it works out, please.

Mike.

[watchdog]

Chrome in portable format does not correctly work in ScPup64-2202: the webcam support is broken. A classic install using the rpm package from google and a little adapting of the desktop file using run-as-spot and apulse1 from ubuntu 22.04 instead correctly works and I can use an external uvc webcam and microphone (my laptop's webcam is upside down and the problem is unresolvable).

[rockedge]

@watchdog Is the internal webcam physically upside down or the stream is rendered upside down?

What stream protocol is being generated mjpeg or something like h264?

It sounds like an interesting problem....what makes it not resolvable?

[watchdog]

@watchdog Is the internal webcam physically upside down or the stream is rendered upside down?

It's phisically upside down: it happens with some Asus laptops of some years ago. Windows had a driver which corrected the stream by software.

It sounds like an interesting problem....what makes it not resolvable?

Read: https://askubuntu.com/questions/749038/ ... ped-chrome

[rockedge]

@watchdog Guvcview in the Video Controls section has an Invert feature that will take an upside down stream and spin it, but I would have to look into a permant way to do the inversion tomake it work with other media software.

Zoneminder could do it and produce an RTSP stream that other apps can capture, but that sounds like extreme overkill.

Possibly using the ZM idea, instead stream the webcam with VLC and use that invert tool. Might be able to capture the stream in a 3rd party app.

[mikewalsh]

Chrome in portable format does not correctly work in ScPup64-2202: the webcam support is broken. A classic install using the rpm package from google and a little adapting of the desktop file using run-as-spot and apulse1 from ubuntu 22.04 instead correctly works and I can use an external uvc webcam and microphone (my laptop's webcam is upside down and the problem is unresolvable).

Hardly surprising, TBH. Slackware being one of the oldest distros still extant, it sticks with tradition, and insists on everything being done in the accepted, correct, time-honoured 'official' way.....and bitches like a good'un if you try to circumvent procedure.

'Slackers' over at LQ.org don't like me. They think there's summat wrong with me, for trying to make things easy-to-use... Talk about having your head stuck in the sand!!

Mike.

[user1111]

Slackware being one of the oldest distros still extant, it sticks with tradition, and insists on everything being done in the accepted, correct, time-honoured 'official' way.....and bitches like a good'un if you try to circumvent procedure.

'Slackers' over at LQ.org don't like me. They think there's summat wrong with me, for trying to make things easy-to-use... Talk about having your head stuck in the sand!!

Thank goodness. They likely opine some having their head stuck up their arse in suggesting that slacko should also become more Windows like. Concepts such as SystemD, PulseAudio and other Windowfication like 'simplification' goes against the Unix philosophy that slackers prefer and will naturally attract contempt if you insist that those that follow such philosophy should do things the incorrect, unproper way.

If you prefer such Windows like simplicity then install and run Windows. Easy. Sadly many of the nix systems have been bribed/suaded to adopt more Windows like methods. Similar to VHS vs Betamax, where vastly technically superior betamax was destroyed by technically inferior VHS marketing.

[williams2]

Attached: screenshot of mpv playing a video, and another instance of mpv playing a video through a vertical flip filter. I have a horizontal flip using the F key, like this: (in /root/.config/mpv/input.conf)
shift+f vf toggle hflip
shift+h vf toggle vflip

(I added the shift+h vf toggle vflip just now, just for this demo.)

This should play the webcam, press H to employ the vflip filter, assuming it has been configured in /root/.config/mpv/input.conf. (in Linux everything is a file ... actually in Linux, everything is a streaming file)

invert.jpg (132.67 KiB) Viewed 9097 times

[mikewalsh]

@user1111 :-

I hold my hand up to that, I guess. I've got too many years of Windows use in my track record to ever really reject making things easy. After all, I came to Puppy partly to get away from the mainstream 'silliness' that insists everything should be done in the terminal.

if you like the traditional way of doing things, fair play to you. I've no problem with that at all - why should I?.....but it's NOT for me. I don't mind using the terminal to troubleshoot, and try stuff out/experiment, but running everything that way?

Nah.

With all the hardware upgrades I've done to this new rig over the course of the pandemic, trying to keep things small/save as much space as possible is something of a 'moot point' nowadays. It no longer even makes sense. Resources are there to be used, not admired then carefully squirrelled-away and left to one side....

We'll "agree to disagree" on this one, I think! 'Slackers' have something of a proprietary attitude to LQ.org.....as though it's "their" site. After all, it IS the official "home" of Slackware, as we all know.....

(*shrug*)

And as for systemd, pulseaudio et al? Give me some credit. I may like certain aspects of Windows behaviour (though not many!), but that's going TOO far. They're all things I absolutely loathe..!

Give me ALSA & traditional init services any day. Systemd is expanding to such an extent, it's trying to take over control of everything. Poettering needs his ass kicked. HARD. If anyone is ignoring Unix/POSIX tradition/compatibility/standards, it's him. Speed up kernel development.....oh, and BTW, ignore POSIX while you're at it?

Jeez.

Mike.

[watchdog]

I have not problems to use my upside down webcam of the asus laptop in any other app: only chrome-chromium-electron are affected by the bug. I use scripts to launch apps prepending:

Code: Select all

export LD_PRELOAD=/usr/lib64/libv4l/v4l1compat.so

That resolves the problem of the upside down webcam.

[Phil_54]

I'm running KLV_Airedale version 17 from an SD card on my Toshiba chromebook which only has galliumOS on SSD. Using chrome-portable I can hear no sound on videos etc. but if I switch to iron-portable, sound works fine. Note that microphone and video are working. Also other distros run fine from SDcard, but I have not tried chrome-portable on them for sound.
I have searched generally for Chrome audio not working, and followed all the pavucontrol/pulseaudio suggestions for fixing (mostly ubuntu sites), but nothing worked. I have chrome (not portable) on my main distro, GalliumOS, and everything is fine. Any hints? I can continue with iron-portable for now, but chrome-portable should work too. Thanks.

[wiak]

I'm running KLV_Airedale version 17 from an SD card on my Toshiba chromebook which only has galliumOS on SSD. Using chrome-portable I can hear no sound on videos etc. but if I switch to iron-portable, sound works fine. Note that microphone and video are working. Also other distros run fine from SDcard, but I have not tried chrome-portable on them for sound.
I have searched generally for Chrome audio not working, and followed all the pavucontrol/pulseaudio suggestions for fixing (mostly ubuntu sites), but nothing worked. I have chrome (not portable) on my main distro, GalliumOS, and everything is fine. Any hints? I can continue with iron-portable for now, but chrome-portable should work too. Thanks.

EDIT2: I just installed KLV-airedale beta17 and chrome-portable worked for me including sound. Did you remember to start it with the LAUNCH app, which makes it correctly run via user spot?

The fact sound works for Iron means your KLV-airedale sound system is all fine. That means something in chrome-portable is breaking it. Most traditional Pups didn't include pulseaudio, which is basically needed for Chrome to work nowadays. I remember that the old Puppy workaround was to use a kind of pseudo pulseaudio/alsa hack called apulse. Only mikewalsh probably knows, but I suspect that chrome portable may include apulse and that's not going to be a good idea when your distro has official pulseaudio anyway all set up and working, or maybe the portable includes actual pulseaudio+configs which is over-writing the working KLV-airedale one(?). Basically you don't want any pulseaudio or apulse in that portable at all. If that's the problem you'd be best with a different chrome-portable build that didn't mess with pulse. My own feeling is that best way of making portable apps to avoid such issues would be to make a separate portable-pulseaudio app should it be required for some distros (e.g. some Pups) but otherwise leave pulseaudio setup to the main distro itself (i.e. for KLV-airedale any portable-pulseaudio app would not be used).

Having said that, I'm guessing what the issue might be! It's interesting that Iron-portable works fine, if that was also designed for traditional pups though - but must be some difference in the way they are put together.

EDIT: But no, I had a quick look inside the chrome-portable and at a glance see no pulseaudio components so will have to leave that one up to mikewalsh to answer. I may try it in KLV-airedale myself later if Mike doesn't get to the bottom of the issue though. But I expect he will suss the problem out. I do note some older pups mentioned... so maybe chrome-portable is only designed for working straight with alsa though I have a recollection that should also be able to get alsa output apps to work through pulseaudio. Maybe it has just been set up to work with Pups and not KLV.

[fredx181]

Solution for me with Mike's chrome-portable on KLV-Airedale is to modify second last line in the LAUNCH script to;
"$HERE/chrome/chrome" --no-sandbox --user-data-dir=$HERE/PROFILE/ --disable-infobars "$@"
Removed 'run-as-spot' and added '--no-sandbox' (this way it runs as root)
I think the no-sound problem is because when running as spot, pulseaudio (as -spot) isn't started properly, but correct me if I'm wrong please, didn't really investigate this)

[wiak]

Solution for me with Mike's chrome-portable on KLV-Airedale is to modify second last line in the LAUNCH script to;
"$HERE/chrome/chrome" --no-sandbox --user-data-dir=$HERE/PROFILE/ --disable-infobars "$@"
Removed 'run-as-spot' and added '--no-sandbox' (this way it runs as root)
I think the no-sound problem is because when running as spot, pulseaudio (as -spot) isn't started properly, but correct me if I'm wrong please, didn't really investigate this)

I modified nothing Fred, and found it worked for me as is.

[fredx181]

Solution for me with Mike's chrome-portable on KLV-Airedale is to modify second last line in the LAUNCH script to;
"$HERE/chrome/chrome" --no-sandbox --user-data-dir=$HERE/PROFILE/ --disable-infobars "$@"
Removed 'run-as-spot' and added '--no-sandbox' (this way it runs as root)
I think the no-sound problem is because when running as spot, pulseaudio (as -spot) isn't started properly, but correct me if I'm wrong please, didn't really investigate this)

I modified nothing Fred, and found it worked for me as is.

Forgot to mention earlier that it doesn't run anyway with default LAUNCH (so could not even test possible sound problem), output:

Code: Select all

root# '/run/media/root/Air/Google_Chrome-portable/LAUNCH' 
cat: '/run/media/root/Air/Google_Chrome-portable/PROFILE/spot/google-chrome/Last Version': No such file or directory
/run/media/root/Air/Google_Chrome-portable/LAUNCH: line 12: [: 104.0.5112.101: unary operator expected
grep: /etc/DISTRO_SPECS: No such file or directory
bash: line 1: cd: /run/media/root/Air/Google_Chrome-portable: Permission denied
bash: line 1: /run/media/root/Air/Google_Chrome-portable/chrome/chrome: Permission denied

Perhaps something went wrong in my KLV-Airedale install, don't know.

Update: Tried now to extract Google_Chrome-portable.tar.gz in /root and works fine running LAUNCH (sound OK too).
That it doesn't work on my nvme partition (labeled Air) may have to do with the way it's mounted by gvfs:
/dev/nvme0n1p3 on /run/media/root/Air type ext4 (rw,nosuid,nodev,relatime,errors=remount-ro,uhelper=udisks2)
Also I found sometime ago that running my Debian mklive-.... scripts (in chroot) didn't work on this partition mounted with gvfs on KLV-Airedale, I guess it's similar problem.
UPDATE2: Tried also now on my second nvme partition, mounted not with gvfs, but by the init script (as it's KLV-Airedale is booted from nvme0n1p2 for me), output from mount :
/dev/nvme0n1p2 on /mnt/layers/uc_ro type ext4 (rw,relatime)
On this partition Google_Chrome-portable works OK, and note the difference of the mount output.
So my conclusion is that I don't really like how gvfs mount handles mounting
(edit: looks like this cannot be fixed, gvfs mount doesn't seem configurable to change mount options, but not sure)

[Phil_54]

Thanks wiak, fred.
I am using LAUNCH to launch. In spot. I only had one instance on my SDcard of the portable, located in a general access folder in same partition.
I downloaded fresh Chrome-portable to spot/Downloads. Tried same youtube video. No sound. Edited penultimate line in LAUNCH as per @fredx181 . Relaunched, got all the stability and security warnings , tried same video and sound works.
I'm reluctant to follow the mount comments as I'm out of my depth of understanding. Not happy with security warnings, although you guys might persuade me not to worry, but iron works without changing anything, so may stick to that for now.
Any advice on best chromium based browser that doesn't take up excessive space, welcome too.

[mikewalsh]

Heh.

It's something of a 'dog's dinner', if I'm honest.

We found quite some time ago that Chrome will still RUN as root, albeit with security warnings and a complete lack of sandboxing. The use of 'spot' allows emulation of the standard /home/user setup as employed by all mainline distros, and is what Chrome is built to expect; like this, sandboxing is as it should be. Brave & Ungoogled Chromium appear to be the same.

Iron, of course, runs as 'root' anyway, so never complains......but sandboxing is non-existent, with "--no-sandbox" in use. The same goes for Slimjet, Vivaldi, Opera, etc., etc.

I find that some 'clones' follow the Chromium code-base rigorously, throwing up dire warnings at the slightest deviation from the expected norm. Others have been modified/recompiled to not be as aggressive about this stuff.

The

Code: Select all

grep -q 'xxxxxxxx' /etc/DISTRO_SPECS

.....is all about assigning the correct onboard version of 'ar' during the upgrade process. 'Ar' was chosen - as opposed to dpkg-deb - because it's the first utility mentioned by blogs across the community as the preferred choice for unpacking .debs. Most go out of their way to make it clear that dpkg-deb is supposed to be used during package management, rather than anything else.

-------------------------------

Yes, I daresay various chunks really need to be re-written once again. It's nowhere near as tidy as it could be, but then I'm no coder......and although I built the updater, and it seems to work, this stuff is right on the limits of my personal understanding. KLV, of course, is the fly in the ointment here because so much of its structure/paths are alien to normal Puppy layout, so special consideration needs to be taken into account to make allowances for this, I guess. I'm not aware of any one, single 'identifying file' that is common to Dogs, Pups AND Void-based builds, so different query routines would have to be combined into a common 'LAUNCH' script in order to have it work under every possible situation.....

(*shrug*)

Mike.