Proposal. For Browser survival in FreeBSD

Hi,

I am recently having some hard time with Chromium, it has worked for a very long time, now it is unusable.

It is about 2-3 years I am using FreeBSD as my main work system, and I think this is one of the thing that must be improved.

I guess all of you use a Web Browser to work these days. I understand that FreeBSD is mostly a server side OS, but, If we want to have any expectation of anybody deciding to use it on the desktop we must fix the browser issue.

Also, consider, FreeBSD is watching to the embedded world. But in the embedded world things have often a screen and a GUI, this GUI is often a just a web page. What people see of your embedded thingy will be just a screen with a browser window open. So, it is better the browser to work reliably, even if we don't care much about desktop users.

I remember when I started I had some other ugly issues with the browser. Then they were solved and for a while all went fine. Now we are back at the starting point.

I am not the most experienced when it comes to ports and packages but, I want to try to propose a possible solution.

What if we implement a package version of a main browser which is fully-statically-compiled? No dependency on other ports or local libs. The browser will be a big binary containing all its universe.

Then, when times come for a new release of the browser to be released we don't delete the old binary-browser package, we just add the new one. Only on the next-next release the older browser will be removed from the repository. In such a way we will be reasonably sure people never remain without their trusted tool.

bye
n.
 
Not sure this is a www/chromium specific issue. I use chromium that I build from ports and have zero issues. Not to discredit your proposal, I just wanted to add that this may not be a widespread issue.
 
I just remembered these issues may be with Chromium version 80 - I am on 79.xxx. I have actually never had issues with Chromium. I only use it because it is purported to be more secure than Firefox. I do not like some of the privacy issues with Firefox but that's another topic.
 
That's what I get for reading on the Internet 🤣 Seriously, I read a couple of months ago that chromium is far ahead of firefox in terms of general browser security, but again, I don't remember from where or whose opinion it was. I like to stick to mainstream browsers for my online activities that involve banking or logins, and for searching and research, I use netsurf. Mainstream to me means one of 2: chromium or Firefox, not sure what else there is for us...
 
I guess all of you use a Web Browser to work these days.
Yes, but not on FreeBSD. Not on Linux either. I only use web browsers on OSes that are actually designed around being human-facing.

I understand that FreeBSD is mostly a server side OS,
Exactly.

but, If we want to have any expectation of anybody deciding to use it on the desktop we must fix the browser issue.
Who is "we" in this sentence? Personally, I don't want to have that expectation. If it works for people as a desktop system: great. If if doesn't: I don't care at all. But my opinion is irrelevant, I'm just one user of FreeBSD. I don't give large donations to FreeBSD, I don't write much code for it, and I'm not a member of the various steering committees and groups.

Also, consider, FreeBSD is watching to the embedded world. But in the embedded world things have often a screen and a GUI, this GUI is often a just a web page.
Few embedded things actually have a screen; most of the interaction with them is done from people's desktop, running against a web server (not browser) on the embedded thing.
But a few embedded things actually do have screens. I would think that anyone who runs an unhardened, consumer-grade browser (whether hardened by statically linking or not) as a UI for a high-reliability embedded thing is not building a good embedded thing, they are building a toy. There are user interface toolkits for the embedded world, and they are not just running a full browser.

I am not the most experienced when it comes to ports and packages but, I want to try to propose a possible solution.
I think the correct solution is to attract many volunteers who want to maintain not only the browser package itself, but also all the packages that the browser depends on. And turn it into a coherent and well-supported set of packages.

And I think in FreeBSD that's just not going to happen. FreeBSD does not make any money from people using it as a desktop, and the FreeBSD developers I know of are not interested in desktop usage. This is very different from Linux: while the biggest source of commercial funding for Linux comes from server systems (RHEL and SUSE support for myriads of servers), the various well-funded Linux organizations explicitly target desktop usage. Just as one example: If you look at the release notes for the latest version of Raspbian (the Debian version that's pre-built for the Raspberry Pi), nearly all of the improvements the Raspian team made are for the GUI. It
seems that the RPi has transitioned from being a teaching system and embedded system to being a low-cost desktop, in the eyes of the people who do OS maintenance for it.

I'm not saying that what your are proposing is a bad idea (except for the technical details, statically linking seems like a step backwards); I'm just saying that within the FreeBSD ecosystem it is likely to get any attention, and other ecosystems are better suited for it.
 
I am recently having some hard time with Chromium, it has worked for a very long time, now it is unusable.

It is about 2-3 years I am using FreeBSD as my main work system, and I think this is one of the thing that must be improved.
What must be improved? You're being extremely vague here and well... it would help if you'd be more specific about the issue. For all I know, and to be honest I'm leaning towards this conclusion, you made a mistake during the upgrade procedure somewhere, ran into issues, and now you're blaming something else for it.

Also... considering the fact that the forum hasn't seen a massive increase of complaints regarding the upgrade I don't see what should be improved here.

I guess all of you use a Web Browser to work these days. I understand that FreeBSD is mostly a server side OS, but, If we want to have any expectation of anybody deciding to use it on the desktop we must fix the browser issue.
What "browser issue"? You have problems with 1 out of a dozen browsers and now the whole OS has a "browser issue"? I beg to differ, I've had no issues what so ever as of late. As such I fully fail to understand what needs to be fixed here.

Still, as always with these things: FreeBSD provides all the tools and info to work on the OS itself, so... maybe start there?

What if we implement a package version of a main browser which is fully-statically-compiled? No dependency on other ports or local libs. The browser will be a big binary containing all its universe.
What makes you so sure that this doesn't exist already?
 
Yeah, I thought about that. More of a privacy issue really. Firefox IS a shorter build using ports. There are a lot of settings in Firefox where data gets sent back to Mozilla and I always turn those off. Of course there are no settings like that in Chromium, but that doesn't mean data isn't getting sent back to the great Google monster.

It's pretty difficult to have a seamless browsing experience without using a big name browser. For us, there are only 2, so it's a dice roll...
 
Yeah, I thought about that. More of a privacy issue really. Firefox IS a shorter build using ports. There are a lot of settings in Firefox where data gets sent back to Mozilla and I always turn those off. Of course there are no settings like that in Chromium, but that doesn't mean data isn't getting sent back to the great Google monster.

It's pretty difficult to have a seamless browsing experience without using a big name browser. For us, there are only 2, so it's a dice roll...

I am torn. There is a lot of sketchy stuff to Firefox but Google is not known for browsing privacy at all. I want to switch back to Chromium on FreeBSD so I don't have to mess with my fontconfig to get fonts to work in Firefox. Windows is the opposite. After all these years Chrome still has that pale, ugly, washed out font.
 
I feel exactly the same way. I wish I could find the article (dev mailing list?) I found about comparing the technologies behind both browsers. Made it sound like FF was ancient on the backend and chromium was more modern, but who knows, I am not a dev. I use no add-ins/plugins on any browser because despite the privacy benefits, any add-in can read all of your browsing traffic. I know some are trustworthy, or rather folks trust some, but I am a bit extreme in that I trust none. I use the 2 browser method of one for logins and one for general surfing.

I have great fonts in both browsers. I normally use "ubuntu" font in both, but that choice doesn't rally matter. What matters for me is to disable bitmap fonts completely, then everything looks sharp and clear.
 
I guess all of you use a Web Browser to work these days.

Don't be so sure. Many people do *not* subscribe to cloud services and so really do not have these problems. Work has continued like always ;)

You are right though, consumer web browsers are all pretty terrible. Think of them as toys and don't depend too much on them.

If you really cannot sever your reliance on internet browsing, consider running a browser-only VM running the most "average" setup of Windows 10 and Chrome.

I do like your proposal however, if anything it would be useful as an experiment to see if we can really clean one up and tailor it to the OS rather than always being dragged along to the latest but weakly tested release.
 
Not trying to derail this thread too much, but I absolutely depend on a browser, for both my job and for general life. I use cloud all the time - on FreeBSD, it's all done through a browser, on my MacBook, via whatever their connection mechanisms are, some browser, some apps.

I like off-shoot browsers, but my fear with these is they may or may not receive attention in terms of maintenance and security fixes so I am not sure I trust them. Other off-shoots are just terrible in terms of rendering which obviously ruins the experience or make some sites unusable.
 
Not sure this is a www/chromium specific issue. I use chromium that I build from ports and have zero issues. Not to discredit your proposal, I just wanted to add that this may not be a widespread issue.

I use almost only packages Sevendogsbsd . So probably we are not on the same version of the application.
Code:
$> chrome --version
Chromium 79.0.3945.130

My FreeBSD runs in a very limited machine, so, when things don't work just well I see it immediately.
 
For Firefox we have the ESR version: www/firefox-esr which more or less serves such purpose.
Maybe such approach is more reasonable?

Honestly, I tried switching from Firefox to Chromium several times, and every time I got bad experience with stability and consistency.


Hi aragats, I have a lot of simpathy for Firefox ESR. I used more Chrome|ium in the past because (1) more people use it (2) the developer console is a bit better IMO (3) pages run into separate processes => If one window goes nut it is improbable that you need to kill all the other 30 windows (4) it seems GoogleDocuments works better in the Chrome family (not sure here). [just by heart, check stuff to be sure]

For me it would be fine if we decide that our "rock-solid-browser" package is based on Firefox. Totally fine. It is not a matter of brand. I require only that it is a known product because I develop stuff for people to use it;)

AFAIK Firefox ESR has a long term support so, Mozilla ensures bugs will be fixed but things will not change and it will be supported for about 5 years [i don't check, right by heart]. Ok, but still we need a way to prevent our package system to mess with the "rock-solid-browser".

What I would like to avoid is that after a pkg upgrade I can't run the browser anymore.

I can accept it for dolphin, okular and other Qt stuff, they are broken so often that I got used to it. But the browser is too much a critic piece of tech. I need it to ask for help or to dig out some kid of solution.
 
What must be improved?
Well I am a SeaMonkey user and portmgr yanked my browser. So how about not yanking ports that are commonly used for starters:
Upstream has poor history of delivering security fixes on time.
2.49.4 was released almost 1 year ago. While 2.49.5 with 60.2
backports is planned[1] it's at least 1 month away while ESR60
will reach EOL in 2 months. By the time 2.57.0 arrives it'll
also be vulnerable.

I don't agree with any part of the comment that was used to remove the port.
The port has been updated several times since removed so this was a bad call to me.

Please give me enough rope to kill myself.
I am able to contain risk and don't need a net nanny.

I totally agree with the original posters sediment in this thread.
When you have a near perfect system and 'upgrading' changes it, the results can be devastating.
I mean a browser is the most simple element I require to use FreeBSD as a desktop.
Xfce changing to GTK3 and was tough to swallow but GTK2 was quite old. That I can deal with.

I have since figured out a way around Seamonkeys removal by using pkg add and using the old packages and all its dependencies.
There is no way new users would be able to pull it off. Heck just locking dependent packages has not worked very well.

Sorry to vent but I have been using the same browser format since 1994 and don't plan on changing because of somebodies opinion of browser security.

I like off-shoot browsers
I too tried to migrate. Falkon, Otter Browser, Iridium, Netsurf. Heck even Firefox ESR is nowhere near SeaMonkey for me.
I guess I am engrained in my ways. If a layout works, why change it. That is my gripe.
I don't want tabs or hidden menus. Just a plain browser that does not change with the wind.

If I could static compile SeaMonkey I would. Unfortunately that ship has sailed log ago.
Most ports won't static compile without work..

I came to FreeBSD from an EOL'ed WinXP and SeaMonkey in ports made the transition seamless.
Having the same OS for browsing and serving has made FreeBSD the only alternative I have used.
Yanking SeaMonkey almost made me leave. I use the browser on FreeBSD daily.

So in summary what we need is stability with the browsers.
I say this not to bash our great OS but for constructive criticism.
 
What must be improved? You're being extremely vague here and well... it would help if you'd be more specific about the issue. For all I know, and to be honest I'm leaning towards this conclusion, you made a mistake during the upgrade procedure somewhere, ran into issues, and now you're blaming something else for it.

Also... considering the fact that the forum hasn't seen a massive increase of complaints regarding the upgrade I don't see what should be improved here.


What "browser issue"? You have problems with 1 out of a dozen browsers and now the whole OS has a "browser issue"? I beg to differ, I've had no issues what so ever as of late. As such I fully fail to understand what needs to be fixed here.

Still, as always with these things: FreeBSD provides all the tools and info to work on the OS itself, so... maybe start there?


What makes you so sure that this doesn't exist already?

Your attitude is entirely critic, I have opened another thread on the forum to discuss this problem. As I have opened a Bugzilla report.

If you don't have the problem good for you.

If you know of a fully static build just let me know. I am not sure of anything.
 
Yes, but not on FreeBSD. Not on Linux either. I only use web browsers on OSes that are actually designed around being human-facing.


Exactly.


Who is "we" in this sentence? Personally, I don't want to have that expectation. If it works for people as a desktop system: great. If if doesn't: I don't care at all. But my opinion is irrelevant, I'm just one user of FreeBSD. I don't give large donations to FreeBSD, I don't write much code for it, and I'm not a member of the various steering committees and groups.


Few embedded things actually have a screen; most of the interaction with them is done from people's desktop, running against a web server (not browser) on the embedded thing.
But a few embedded things actually do have screens. I would think that anyone who runs an unhardened, consumer-grade browser (whether hardened by statically linking or not) as a UI for a high-reliability embedded thing is not building a good embedded thing, they are building a toy. There are user interface toolkits for the embedded world, and they are not just running a full browser.


I think the correct solution is to attract many volunteers who want to maintain not only the browser package itself, but also all the packages that the browser depends on. And turn it into a coherent and well-supported set of packages.

And I think in FreeBSD that's just not going to happen. FreeBSD does not make any money from people using it as a desktop, and the FreeBSD developers I know of are not interested in desktop usage. This is very different from Linux: while the biggest source of commercial funding for Linux comes from server systems (RHEL and SUSE support for myriads of servers), the various well-funded Linux organizations explicitly target desktop usage. Just as one example: If you look at the release notes for the latest version of Raspbian (the Debian version that's pre-built for the Raspberry Pi), nearly all of the improvements the Raspian team made are for the GUI. It
seems that the RPi has transitioned from being a teaching system and embedded system to being a low-cost desktop, in the eyes of the people who do OS maintenance for it.

I'm not saying that what your are proposing is a bad idea (except for the technical details, statically linking seems like a step backwards); I'm just saying that within the FreeBSD ecosystem it is likely to get any attention, and other ecosystems are better suited for it.

Hi ralphbsz , i respect your point of view but i can't share it. If you don't use FBS as desktop you can't feel the pain. I use FBS to develop almost everything except Android stuff. Well, sometimes things go wrong and it is not going to be nice. The worst is, when also the browser leaves you. Then, you just need another computer to fix your computer.

About "we". If was good at Makefiles, was a Pudriere person, had strong hardware to compile a browser just for fun, and if I was a C++ developer, at least 1 month per year, I would have not even opened this thread. I would have just solved my problem. Since this is not the case. I can just raise the issue and hope somebody in the community who is good with the required tech is willing to solve it. Too much out of my scope.

I would respectfully disagree with you on a few points which are a bit out of the main topic.

1] embedded can be controlled via web or App but that not always is desirable. eg-0, think of your car dashboard. eg-1 Think mesurement instruments, you just don't want to use the Phone, and follow a super boring procedure to read that damn temperature. eg-2. Ever tried an headless oscilloscope, well, it is kind of annying, if you can buy one with a screen you will just do it.

In general, if you have your device located in a room and you can add the basic controls over it, the user interaction will be easier and everybody will be happier. Then, computer guys can always use in order: (1) the App (2) the web interface (3) ssh (4) the serial cable. But for others, all these things are in order of "annoyment" and "detestability".

2] interfaces. I wrote a few weeks ago a little program in PySide2, guess what, there is a browser there. It is not called Chrome but I think the engine is the same [by heart, check to be sure]. Then if you try Google something like "map widget" you will find, as i did, that the recommended solution is to just run the Web widget and see the map inside it. So the trend is pretty much clear, Interfaces are converging to the web. Do i like it? Not much. But I can live with that.

3] static linking. I understand that may not be fashionable. But that is the best I can figure out to actually improve the status quo. The status quo being, if you run "pkg install foobarbaz" something else may be upgraded by necessity and the consequence being, your browser not working anymore. I realize it is not an elegant way to solve things but, just for the browser we may close one eye ... isn't it the future OS ?
 
I too tried to migrate. Falkon, Otter Browser, Iridium, Netsurf. Heck even Firefox ESR is nowhere near SeaMonkey for me.
What about Seamonkey makes it so attractive to you? Familiarity, perhaps? Put a different way, what is lacking in other implementations you've tried?

Have you given www/opera a go? I don't want to presume your reasoning, but Opera boasts a feature-set more akin to Seamonkey's "communication suite" rather than being a simple web browser, so you might find it to your liking after configuring it as you wish. And to be completely fair, it's the only "suite" I'm aware of; such software is a bit of a dying breed as just getting HTML+CSS+JS to all work correctly is quite a task already these days.

Edit: Nevermind. Even Opera's own site doesn't seem to function quite correctly in it, and it's apparently version 12.x when the newest is 66.x... Perhaps this is why you're stuck with Seamonkey: nothing else offers the complete package, and mucking about with the settings for different pieces of software is a pain compared to the cohesive configuration experience that Seamonkey offers?
 
Not trying to derail this thread too much, but I absolutely depend on a browser, for both my job and for general life. I use cloud all the time - on FreeBSD, it's all done through a browser, on my MacBook, via whatever their connection mechanisms are, some browser, some apps.

Totally agree here. I don't own a TV since about 12 years ! I just watch stuff on the computer, mostly browser these days: youtube and fubo.tv. Since a long time chat and phone calls have been through the web for me, via Whatsapp. At work without a browser would be simply unthinkable, also our timesheet is via web. I am totally browser dependent.

It is true, I run in a VM, so when I am really in trouble I use the browser in MacOS, but I try do to it less and less, because the system I love is FreeBSD, where almost everything has been configured to work the way a like :)
 
Don't be so sure. Many people do *not* subscribe to cloud services and so really do not have these problems. Work has continued like always ;)
In my current workplace, a big one, everything has a browser procedure, I have a GoogleDocument just for the list of this "important" links. So, I guess my view is biased.

If you really cannot sever your reliance on internet browsing, consider running a browser-only VM running the most "average" setup of Windows 10 and Chrome.
I considered something similar but still I have not found the silver bullet. I am alredy in a VM, so my only reasonable choice is to run the browser on metal and "import" its window in FreeBSD in some way.

I do like your proposal however, if anything it would be useful as an experiment to see if we can really clean one up and tailor it to the OS rather than always being dragged along to the latest but weakly tested release.
This is exactly my point !
 
Made it sound like FF was ancient on the backend and chromium was more modern, but who knows, I am not a dev.

The engine of Firefox has changed around version 60, resulting in a dramatic performance increase.
Thunderbird has adopted the same engine in version 68.

As to privacy, Firefox asks you on its first launch whether you want to contribute to the project by sending telemetry.
You can still change your mind afterwards, the settings are in Preferences.
Nothing is hidden.
 
Well I am a SeaMonkey user and portmgr yanked my browser. So how about not yanking ports that are commonly used for starters:
I'm a SeaMonkey user too, have been for many years, and my update never gave me problems. I'm using Portmaster btw, so a pretty vanilla'ish upgrade scheme.
 
Back
Top