WebKit Bugzilla
New
Browse
Log In
×
Sign in with GitHub
or
Remember my login
Create Account
·
Forgot Password
Forgotten password account recovery
RESOLVED INVALID
256471
Minibrowser starts the network process but the web process crashes
https://bugs.webkit.org/show_bug.cgi?id=256471
Summary
Minibrowser starts the network process but the web process crashes
Mero
Reported
2023-05-08 09:45:28 PDT
The browser starts without loading the default webkit page. When I enter the address and press enter, it prints out the message that the web process crashed. I can see the network process running. A plenty of messages pop like Gtk-WARNING **: 16:06:33.489: Theme parsing error: lightdm-gtk-greeter.css:91:16: not a number Gtk-WARNING **: 16:06:33.489: Theme parsing error: lightdm-gtk-greeter.css:91:16: Expected a string. I suppose it cannot find atk, dbus, gtk or glib configuration and resorts to the system ones which are outdated. warning if I click preferences in automation mode Gtk-WARNING **: 16:26:07.243: Could not load a pixbuf from /org/gtk/libgtk/theme/Adwaita/assets/check-symbolic.svg. This may indicate that pixbuf loaders or the mime database could not be found.
Attachments
Add attachment
proposed patch, testcase, etc.
Michael Catanzaro
Comment 1
2023-05-08 12:21:28 PDT
Please attach a backtrace for the crashing process. Follow the instructions at
https://blogs.gnome.org/mcatanzaro/2021/09/18/creating-quality-backtraces-for-crash-reports/
if you don't know how. There's nothing we can do about a crash if you do not provide a backtrace. The CSS errors and missing icon should not cause a crash.
Mero
Comment 2
2023-05-08 15:34:59 PDT
Multiple errors of this kind are reported from kernel kernel: [915620.849021] WebKitWebProces[24131]: segfault at 8 ip 00007fae013d9b85 sp 00007fffb373f540 error 4 in libwebkit2gtk-4.0.so.37.56.11[7fae011b5000+192e000] The packages you recommend in the website are not available in the system or not operating. E.g. apport wont report if the minibrowser is not set up in the system, while I build it from source and thus it is not helpful. You are saying in the website that CSS rendering issues are caused by memory corruption. And if run in a low RAM setting is it possible that this is the issue.
Mero
Comment 3
2023-05-09 03:22:37 PDT
Strace minibrowser prints multiple lines recvmsg(3, 0x7fffe2945030, 0)= -1 EAGAIN (Resource temporarily unavailable) Strace on the network process prints nothing. So I suppose it is a file or driver failing to run or detect.
Michael Catanzaro
Comment 4
2023-05-09 05:10:32 PDT
Sorry, without a backtrace we will not investigate this. Feel free to reopen if you can get a backtrace. You'll need to disable apport. (In reply to Mero from
comment #2
)
> You are saying in the website that CSS rendering issues are caused by memory > corruption. And if run in a low RAM setting is it possible that this is the > issue.
No, I'm not saying either of those things. I'm saying WebKit is crashing and we need a backtrace to see why. (In reply to Mero from
comment #3
)
> Strace minibrowser prints multiple lines > > recvmsg(3, 0x7fffe2945030, 0)= -1 EAGAIN (Resource temporarily unavailable) > > Strace on the network process prints nothing. > > So I suppose it is a file or driver failing to run or detect.
No, that's harmless.
Mero
Comment 5
2023-05-09 06:12:32 PDT
Theme parsing error: lightdm-gtk-greeter.css disappears if I switch the theme to Adwaita. So I should have built the font themes. They are not compatible. I disable gobject introspection and wayland in the build since they require new qemu and headers. I'm not sure if the streamers, gtk and webkit cross-depend on them too. Still, when I run cmake, it doesn't stop the build, so this may not be the issue. Backtraces are not available to install. In latest releases they may help but then I will install the newly released OS and not need them at all or new builds. Valgrind changes the system to i386 and not sure if it can help at all since if it is memory corruption nothing can be done. The new software simply will be incompatible with outdated hardware.
Mero
Comment 6
2023-05-09 06:15:11 PDT
(In reply to Michael Catanzaro from
comment #4
)
> Sorry, without a backtrace we will not investigate this. Feel free to reopen > if you can get a backtrace. You'll need to disable apport. > > (In reply to Mero from
comment #2
) > > You are saying in the website that CSS rendering issues are caused by memory > > corruption. And if run in a low RAM setting is it possible that this is the > > issue. > > No, I'm not saying either of those things. I'm saying WebKit is crashing and > we need a backtrace to see why. > > (In reply to Mero from
comment #3
) > > Strace minibrowser prints multiple lines > > > > recvmsg(3, 0x7fffe2945030, 0)= -1 EAGAIN (Resource temporarily unavailable) > > > > Strace on the network process prints nothing. > > > > So I suppose it is a file or driver failing to run or detect. > > No, that's harmless.
Strace does nothing to help. This missing resource may be the web process which crashes.
Mero
Comment 7
2023-05-09 07:07:16 PDT
My idea and suggestion is that the build is useless for older hardware and these debugging tools are useless completely. The webkit better change the whole architecture and build all packages needed for webkit with static libraries. So that we as users dont need to look how it can be built. And in C language since C++ changes continuously changes and has very high requirements for RAM. The example is cmake which is successful and builds all it needs.
Mero
Comment 8
2023-05-09 07:23:08 PDT
Is it good if I can upload the libraries and binaries somewhere and you will debug it?
Michael Catanzaro
Comment 9
2023-05-09 12:13:18 PDT
No, we will only look at a backtrace taken with gdb.
Mero
Comment 10
2023-05-10 02:41:45 PDT
Ok but how will be made the backtrace? I can not do it since I can not download the required packages. Even if it is possible, you are saying that they will take 20 GB disk space or more so it is impossible to me. If I build it as a debug build it will take ages and will crash RAM and disk drive, maybe 100 GB. The Gcc compiler is not good at latest C++ and requires a lot of memory. Clang is doubtful and yet can not build glibc. I hardly manage to build the release and it crashes. In short, it is either impossible or pointless to me.
Mero
Comment 11
2023-05-10 02:48:44 PDT
By the way why dont you make the github package downloadable as a small package with the build scripts added? It has useful scripts to build all libraries and flatpak, but currently downloads a huge 10 GB zip, which is not needed and it is impossible to download from the github because download stops or is killed before downloading. I can download only the release but there are no scripts in it.
Carlos Alberto Lopez Perez
Comment 12
2023-05-10 13:13:06 PDT
Those binaries are intended to run inside the flatpak environment. They won't run on any other distro or outside of the flatpak environment.
Mero
Comment 13
2023-05-11 02:03:25 PDT
(In reply to Carlos Alberto Lopez Perez from
comment #12
)
> Those binaries are intended to run inside the flatpak environment. They > won't run on any other distro or outside of the flatpak environment.
That is what I mean too. It may be this dependency package flatpak that I didn't build and it crashes the run now. Or something else needed. I disabled gobject introspection, bubblewrap, and wayland since they need new headers, qemu and maybe other packages. However, cmake didn't stop the minibrowser build and it is impossible to catch all these dependencies. As far as I know, flatpak also may require the latest system headers to build and run. I read the scripts in the github, but I suppose they are only for the mac systems and cannot be applied.
Carlos Alberto Lopez Perez
Comment 14
2023-05-11 10:02:29 PDT
(In reply to Mero from
comment #13
)
> (In reply to Carlos Alberto Lopez Perez from
comment #12
) > > Those binaries are intended to run inside the flatpak environment. They > > won't run on any other distro or outside of the flatpak environment. > > That is what I mean too. It may be this dependency package flatpak that I > didn't build and it crashes the run now. Or something else needed. I > disabled gobject introspection, bubblewrap, and wayland since they need new > headers, qemu and maybe other packages. However, cmake didn't stop the > minibrowser build and it is impossible to catch all these dependencies. As > far as I know, flatpak also may require the latest system headers to build > and run. I read the scripts in the github, but I suppose they are only for > the mac systems and cannot be applied.
Ok. I'm not sure what you are exactly trying to do but it looks to me it is not going to work. I will explain briefly how to build WebKit for Linux (either WPE or WebKitGTK): 1) First about the machine: You need a really beefy machine. WebKit is huge and complex c++ software that takes ages to build. The bare minimum you need is a 4-core Intel CPU with 12GB of RAM and a SSD with at last 50GB free of disk space. With a setup like this one it would take around 4-8 hours to build. If you want to do it in less than an hour then you need at least 16 powerful cores and and 64GB of RAM. The build system will try to use all your cores by default so you need at least 2GB of RAM free per vcore otherwise the compiler will be killed by the system or will fail with strange errors because there is not enough RAM. Anything less than that don't even try. Is not going to work. 2) Then the process of building WebKitGTK is documented here:
https://trac.webkit.org/wiki/BuildingGtk
Notice the first step is running this command: Tools/Scripts/update-webkitgtk-libs That will download and install the flatpak SDK that you need to continue developing. That is a container that has all the required dependencies. So the stes are: 1. Get the flatpak SDK $ Tools/Scripts/update-webkitgtk-libs 2. Build WebKit $ Tools/Scripts/build-webkit --gtk 3. Run the minibrowser: Tools/Scripts/run-minibrowser --gtk If you want to try the WPE port just replace "--gtk" with "--wpe" Hope it helps
Michael Catanzaro
Comment 15
2023-05-11 11:11:17 PDT
The flatpak SDK is really only useful if you are planning to develop WebKit itself. You cannot install WebKit if built that way, so those instructions are not appropriate here. Mero surely wants to use CMake directly.
Carlos Alberto Lopez Perez
Comment 16
2023-05-11 13:29:13 PDT
(In reply to Michael Catanzaro from
comment #15
)
> The flatpak SDK is really only useful if you are planning to develop WebKit > itself. You cannot install WebKit if built that way, so those instructions > are not appropriate here. Mero surely wants to use CMake directly.
To install it on your system you have to build it against the libraries of your system. You can do that but unless you are running an edge distro expect to have to disable several things to be able to build it because it may need newer libraries for such features than what your system ships Another approach is to generate a bundle from the flatpak build with the the tool generate-bundle. You can do that by doing: 1. Build with the flatpak SDK like I described above 2. Run the scrip to generate the bundle like this: $ Tools/Scripts/generate-bundle --platform gtk --bundle all --syslibs bundle-all Then it should generate a .tar.xz file inside the directory WebKitBuild with the bundle 4. Now, in theory, you can unpack the file on your system and execute the MiniBrowser script and it should just work. I have uploaded here a bundle for WebKitGTK built just now at
263979@main
so you can test it. Check it here:
https://people.igalia.com/clopez/wkbug/256471/bundle_webkitgtk_test.tar.xz
And last, there are some other bundles built specifically for Ubuntu-20.04 with some features disabled so it can build for that system (note: this ones only work on ubuntu-20.04 and you need execute the install-deps script inside before running them). See:
https://webkitgtk.org/built-products/x86_64/release/nightly/Ubuntu-20.04/MiniBrowser/
Mero
Comment 17
2023-05-12 09:41:12 PDT
Thank you ! I tried it but it did not run as I expected. I have no a modern pc (mine is a 2009 make:). Well, maybe it may be compared to a modern phone with 2 GB RAM :).And that is why I cannot install latest desktop Linux kernel, drivers, etc. And you understand that if I had it, the build would be pointless and I would simply run all in the latest system. Anyway, what you send is very helpful to check the running packages and most of them are on my build too. I would say it is 80-90 % cover in your and my builds. Let me try with your scripts and I will let you know if I can run it. Certainly, your build depends on modern glibc and it wont run in my system.
Carlos Alberto Lopez Perez
Comment 18
2023-05-12 11:28:35 PDT
(In reply to Mero from
comment #17
)
> Thank you ! I tried it but it did not run as I expected. I have no a modern > pc (mine is a 2009 make:). Well, maybe it may be compared to a modern phone > with 2 GB RAM :).And that is why I cannot install latest desktop Linux > kernel, drivers, etc. And you understand that if I had it, the build would > be pointless and I would simply run all in the latest system. >
That bundle that I created from the binaries of the flatpak-sdk tries to use its own glibc which is also included on the bundle. Actually you can even run it on distros that don't ship glibc like Alpine. However, the version of glibc bundled was built with a requirement of Linux kernel >= 3.2, so as long as you have kernel 3.2 or newer it should work (in theory). If you have an older version of the kernel then it won't work. But that is really really ancient. I'm checking that Linux 3.2 was released in 2012!
> Anyway, what you send is very helpful to check the running packages and most > of them are on my build too. I would say it is 80-90 % cover in your and my > builds. Let me try with your scripts and I will let you know if I can run > it. Certainly, your build depends on modern glibc and it wont run in my > system.
Glad to know that it helped somehow. BTW, in case you find interesting this: - When you invoke the generate-bundle script you can tell it if it should try to bundle everything (even glibc and the graphics drivers) by passing '--syslibs=bundle-all' or if it should try to only bundle what is not shipped in the system (so only webkit and some library built outside of the standard system paths if webkit links with it) and generate a install script for the needed system packages by passing '--syslibs=generate-install-script' (currently that only works on .deb based distros)
Carlos Alberto Lopez Perez
Comment 19
2023-05-12 11:39:17 PDT
(In reply to Carlos Alberto Lopez Perez from
comment #18
)
> (In reply to Mero from
comment #17
) > > Thank you ! I tried it but it did not run as I expected. I have no a modern > > pc (mine is a 2009 make:). Well, maybe it may be compared to a modern phone > > with 2 GB RAM :).And that is why I cannot install latest desktop Linux > > kernel, drivers, etc. And you understand that if I had it, the build would > > be pointless and I would simply run all in the latest system. > > > > That bundle that I created from the binaries of the flatpak-sdk tries to use > its own glibc which is also included on the bundle. Actually you can even > run it on distros that don't ship glibc like Alpine. > > However, the version of glibc bundled was built with a requirement of Linux > kernel >= 3.2, so as long as you have kernel 3.2 or newer it should work (in > theory). If you have an older version of the kernel then it won't work. But > that is really really ancient. I'm checking that Linux 3.2 was released in > 2012!
BTW, If your system doesn't have graphics drivers with proper OpenGL support that is an issue as WebKit tries to use accelerated compositing by default. Maybe you can workaround this exporting the following environment variable that tells WebKit to not use the accelerated compositing paths. $ export WEBKIT_DISABLE_COMPOSITING_MODE=1 $ ./MiniBrowser You can also try to load also the URL 'about:gpu' on the MiniBrowser that will tell you info about the detected capabilities of the graphics drivers
Mero
Comment 20
2023-05-12 12:57:28 PDT
It starts only with this option $ export WEBKIT_DISABLE_COMPOSITING_MODE=1 $ ./MiniBrowser If not included, it leads to Aborted core dumped error and crashes/ Sadly, I also run it in my browser but if fails. I thinks it is a dependency which is not built and cannot run. In your build it starts wonderful and prints this error Invalid TLS Certificate Failed to load
https://www.webkitgtk.org/
: Do you want to continue ignoring the TLS errors? I suppose it is a gnutils error and I click on yes and it loads all. I check on youtube and it cannot load streams. Videos blink and not completely seen. But I think it is wonderful. I need it only for the webpages. If you interested in the stream error, sending it here WebKitWebProcess:7146): GStreamer-CRITICAL **: 22:51:09.730: gst_mini_object_unref: assertion 'GST_MINI_OBJECT_REFCOUNT_VALUE (mini_object) > 0' failed Many thanks Gentlemen !
Carlos Alberto Lopez Perez
Comment 21
2023-05-12 16:36:08 PDT
(In reply to Mero from
comment #20
)
> It starts only with this option > > $ export WEBKIT_DISABLE_COMPOSITING_MODE=1 > $ ./MiniBrowser > > If not included, it leads to Aborted core dumped error and crashes/ Sadly, I > also run it in my browser but if fails. I thinks it is a dependency which is > not built and cannot run. In your build it starts wonderful and prints this > error > > Invalid TLS Certificate > Failed to load
https://www.webkitgtk.org/
: Do you want to continue ignoring > the TLS errors? > > I suppose it is a gnutils error and I click on yes and it loads all. > > I check on youtube and it cannot load streams. Videos blink and not > completely seen. But I think it is wonderful. I need it only for the > webpages. > If you interested in the stream error, sending it here > > WebKitWebProcess:7146): GStreamer-CRITICAL **: 22:51:09.730: > gst_mini_object_unref: assertion 'GST_MINI_OBJECT_REFCOUNT_VALUE > (mini_object) > 0' failed > > Many thanks Gentlemen !
out of curiosity.. which linux distro and version are you running?
Mero
Comment 22
2023-05-13 03:03:09 PDT
(In reply to Carlos Alberto Lopez Perez from
comment #21
)
> (In reply to Mero from
comment #20
) > > It starts only with this option > > > > $ export WEBKIT_DISABLE_COMPOSITING_MODE=1 > > $ ./MiniBrowser > > > > If not included, it leads to Aborted core dumped error and crashes/ Sadly, I > > also run it in my browser but if fails. I thinks it is a dependency which is > > not built and cannot run. In your build it starts wonderful and prints this > > error > > > > Invalid TLS Certificate > > Failed to load
https://www.webkitgtk.org/
: Do you want to continue ignoring > > the TLS errors? > > > > I suppose it is a gnutils error and I click on yes and it loads all. > > > > I check on youtube and it cannot load streams. Videos blink and not > > completely seen. But I think it is wonderful. I need it only for the > > webpages. > > If you interested in the stream error, sending it here > > > > WebKitWebProcess:7146): GStreamer-CRITICAL **: 22:51:09.730: > > gst_mini_object_unref: assertion 'GST_MINI_OBJECT_REFCOUNT_VALUE > > (mini_object) > 0' failed > > > > Many thanks Gentlemen ! > > out of curiosity.. which linux distro and version are you running?
lubuntu 14
Mero
Comment 23
2023-05-13 03:10:11 PDT
(In reply to Carlos Alberto Lopez Perez from
comment #16
)
> (In reply to Michael Catanzaro from
comment #15
) > > The flatpak SDK is really only useful if you are planning to develop WebKit > > itself. You cannot install WebKit if built that way, so those instructions > > are not appropriate here. Mero surely wants to use CMake directly. > > To install it on your system you have to build it against the libraries of > your system. You can do that but unless you are running an edge distro > expect to have to disable several things to be able to build it because it > may need newer libraries for such features than what your system ships > > Another approach is to generate a bundle from the flatpak build with the the > tool generate-bundle. > > You can do that by doing: > > 1. Build with the flatpak SDK like I described above > > 2. Run the scrip to generate the bundle like this: > > $ Tools/Scripts/generate-bundle --platform gtk --bundle all --syslibs > bundle-all > > Then it should generate a .tar.xz file inside the directory WebKitBuild with > the bundle > > 4. Now, in theory, you can unpack the file on your system and execute the > MiniBrowser script and it should just work. > > I have uploaded here a bundle for WebKitGTK built just now at
263979@main
so > you can test it. > Check it here: >
https://people.igalia.com/clopez/wkbug/256471/bundle_webkitgtk_test.tar.xz
> > And last, there are some other bundles built specifically for Ubuntu-20.04 > with some features disabled so it can build for that system (note: this ones > only work on ubuntu-20.04 and you need execute the install-deps script > inside before running them). See: >
https://webkitgtk.org/built-products/x86_64/release/nightly/Ubuntu-20.04/
> MiniBrowser/
That was my request to you and webkit developers if you read our discussion with Michael above. Is it possible to include the build scripts of the libraries in the release build on the webkit website. That saves life since different versions change the requirements and versions and it is very hard to follow them in order to build.Something similar to the build of the cmake package and it downloads or contains the required packages and bootstraps or builds them.
Mero
Comment 24
2023-05-13 03:14:57 PDT
(In reply to Carlos Alberto Lopez Perez from
comment #14
)
> (In reply to Mero from
comment #13
) > > (In reply to Carlos Alberto Lopez Perez from
comment #12
) > > > Those binaries are intended to run inside the flatpak environment. They > > > won't run on any other distro or outside of the flatpak environment. > > > > That is what I mean too. It may be this dependency package flatpak that I > > didn't build and it crashes the run now. Or something else needed. I > > disabled gobject introspection, bubblewrap, and wayland since they need new > > headers, qemu and maybe other packages. However, cmake didn't stop the > > minibrowser build and it is impossible to catch all these dependencies. As > > far as I know, flatpak also may require the latest system headers to build > > and run. I read the scripts in the github, but I suppose they are only for > > the mac systems and cannot be applied. > > Ok. I'm not sure what you are exactly trying to do but it looks to me it is > not going to work. > > I will explain briefly how to build WebKit for Linux (either WPE or > WebKitGTK): > > 1) First about the machine: You need a really beefy machine. WebKit is huge > and complex c++ software that takes ages to build. The bare minimum you need > is a 4-core Intel CPU with 12GB of RAM and a SSD with at last 50GB free of > disk space. With a setup like this one it would take around 4-8 hours to > build. If you want to do it in less than an hour then you need at least 16 > powerful cores and and 64GB of RAM. > > The build system will try to use all your cores by default so you need at > least 2GB of RAM free per vcore otherwise the compiler will be killed by the > system or will fail with strange errors because there is not enough RAM. > Anything less than that don't even try. Is not going to work. > > 2) Then the process of building WebKitGTK is documented here: > >
https://trac.webkit.org/wiki/BuildingGtk
> > Notice the first step is running this command: > Tools/Scripts/update-webkitgtk-libs > > That will download and install the flatpak SDK that you need to continue > developing. > > That is a container that has all the required dependencies. > > > So the stes are: > > 1. Get the flatpak SDK > $ Tools/Scripts/update-webkitgtk-libs > > 2. Build WebKit > $ Tools/Scripts/build-webkit --gtk > > 3. Run the minibrowser: > Tools/Scripts/run-minibrowser --gtk > > > > If you want to try the WPE port just replace "--gtk" with "--wpe" > > Hope it helps
That is big hindrance in building webkit. I think there is something wrong in webkit architecture. I think you agree that I can build all packages needed to webkit with a humble hardware and then finally, the most important package webkit requires such a powerful hardware and it is impossible.
Mero
Comment 25
2023-05-14 01:26:39 PDT
One question to you please. Would you tell how I can run it without writing to the /home/.cache folder. E.g. I may configure it in the config file to write the cache only to one folder on the usb and ban writing to the /home folder or any other system folder. My build is good only for the recycle bin, but yours is excellent and I need it highly.
Carlos Alberto Lopez Perez
Comment 26
2023-05-15 03:15:19 PDT
(In reply to Mero from
comment #25
)
> One question to you please. Would you tell how I can run it without writing > to the /home/.cache folder. E.g. I may configure it in the config file to > write the cache only to one folder on the usb and ban writing to the /home > folder or any other system folder. My build is good only for the recycle > bin, but yours is excellent and I need it highly.
I guess you can use the XDG_CACHE_HOME environment variable - $XDG_CACHE_HOME defines the base directory relative to which user-specific non-essential data files should be stored. If $XDG_CACHE_HOME is either not set or empty, a default equal to $HOME/.cache should be used. More info:
https://specifications.freedesktop.org/basedir-spec/basedir-spec-latest.html
Carlos Alberto Lopez Perez
Comment 27
2023-05-15 03:20:28 PDT
(In reply to Mero from
comment #22
)
> > out of curiosity.. which linux distro and version are you running? > > lubuntu 14
That is very old and we don't support such old stack. I'm actually surprised the bundle managed to start there. I think you would be better served by upgrading your software stack instead of fighting with it. You can install Debian that is specially friendly with old hardware (it needs low resources) and let's you choose which desktop you wish at installation time, so you can choose XFCE or LXDE which should work fine with computers with old hardware or poor graphics drivers. See:
https://wiki.debian.org/DesktopEnvironment#Desktop_Environment
My 2 cents
Mero
Comment 28
2023-05-15 04:17:43 PDT
It can run but cannot be built:) The build is the problem. Linux saves and helps older hardware since Windows and Macs always require a new pc. No choice. I would not like to dispose of this pc, if it is running and trying to only renew or update what I can do, so that it can be operational. If I set up a new Linux, it will also require more RAM for the build and gets fast to the limit. Thank you for your help !
Carlos Alberto Lopez Perez
Comment 29
2023-05-15 06:55:49 PDT
(In reply to Mero from
comment #28
)
> It can run but cannot be built:) The build is the problem. Linux saves and > helps older hardware since Windows and Macs always require a new pc. No > choice. I would not like to dispose of this pc, if it is running and trying > to only renew or update what I can do, so that it can be operational. If I > set up a new Linux, it will also require more RAM for the build and gets > fast to the limit. > > Thank you for your help !
Debian doesn't have to be built and requires low resources. It should run fine on old hardware, is just a matter of installing it. Give it a try. If you want to check if it will run before installing it, then you can try to boot this live CDs
https://cdimage.debian.org/debian-cd/current-live/amd64/iso-hybrid/
(hint: try the lxqt or xfce ones) You can boot them also from an usb stick just by copying raw the ISO file. Example: dd if=file-name.iso of=/dev/sdb bs=1M # assuming /dev/sdb is your USB device Debian also ships the last stable version of WebKitGTK and WPEWebKit
Mero
Comment 30
2023-05-15 12:52:52 PDT
Do you read this
https://cdimage.debian.org/debian-cd/current-live/amd64/iso-hybrid/
>The minimum recommended RAM for using a desktop environment on a live image is >2 GiB. If you have a system with less memory, your system will not work well.
The best is older Ubuntu with RAM ~ 1 GiB. Newer Ubuntu is last resort. Debian desktop requires more resources.
Carlos Alberto Lopez Perez
Comment 31
2023-05-15 13:13:45 PDT
(In reply to Mero from
comment #30
)
> Do you read this > >
https://cdimage.debian.org/debian-cd/current-live/amd64/iso-hybrid/
> >The minimum recommended RAM for using a desktop environment on a live image is >2 GiB. If you have a system with less memory, your system will not work well. > > The best is older Ubuntu with RAM ~ 1 GiB. Newer Ubuntu is last resort. > Debian desktop requires more resources.
I see, this is specially true for the live image since it can't write to the disk and has to store more things than usual on RAM. However if you install it to disk it should be enough with 1GB of RAM if you choose a lightweight desktop like XFCE or LXDE/LXQT Here are the documented system requirements:
https://www.debian.org/releases/jessie/mips/ch03s04.html.en
At any rate, once you start a browser and start loading a few websites 1GB of RAM is likely not enough for comfortable usage. The browser itself can use 1GB or more.
Carlos Alberto Lopez Perez
Comment 32
2023-05-15 16:09:12 PDT
(In reply to Carlos Alberto Lopez Perez from
comment #31
)
> Here are the documented system requirements: >
https://www.debian.org/releases/jessie/mips/ch03s04.html.en
>
That link is outdated, is for Debian 8, the updated one (Debian 11) is
https://www.debian.org/releases/stable/amd64/ch03s04.en.html
Comparing requirements for both versions, I can see your point that newer versions require more RAM.
Mero
Comment 33
2023-05-16 04:21:12 PDT
Ubuntu and Debian are much the same. I work on Ubuntu and will need more time to switch to Debian. Latest Ubuntu and Debian change kernel and thus may be incompatible with older hardware. Well, what is more important is a lightweight browser. And this was the minibrowser. I can not find another one which is easier to build and light-weight to run. Chrome loads a lot of resources and writes history, data etc. Firefox too. I don't need this data and need to stop it writing to the drive. It is very annoying to clean continuously all this unnecessary data. And I need only the basic HTML for the web pages. One that I can check my email on, for example.
Note
You need to
log in
before you can comment on or make changes to this bug.
Top of Page
Format For Printing
XML
Clone This Bug