Porteus USB live install

[M. Eerie]

SUGGESTION: when using old, cheap and slow 2.0 USB stick, go for VFAT LIVE installation.

Indeed. Anyway, it wasn't your script that failed. It was an old stick that I was hoping would work because it contains a small led to indicate I/O operations. When not working with journaling, it would have been a small but great help instead of having to resort to sync or watching with btop if write operations have finished. As for the other one... it was a disaster due to a power outage.

As practically everything in the latter I have duplicated (although scattered) and is a 3.1 one, I have used it to test the usb-install version of your script and it works well. However, I am testing with Nemesis instead of Porteu{s,x}. Although it works relatively well, I don't know exactly where to place the "config" folder. If I put it in the first partition, the cheatcodes seem to be ignored. If I put them in the second partition, I still get multiple errors related to etmpfiles (among others) if I remember correctly.

[robang74]

it would have been a small but great help instead of having to resort to sync or watching with btop if write operations have finished.

My script usb-install (which is also used by net-install) clearly states when the usb is sync and eject it, so you do not need look at anything but just wait.

ERRATA CORRIGE

Old and slow USB sticks can timeout the umount synchronisation timeout and also eject would fail. In this corner case, as long as exist the device in `/proc/partitions`, it means that it cannot be safely disconnected. However, much probably it does not worth using it.

[robang74]

However, I am testing with Nemesis instead of Porteu{s,x}

Who choose that name make a "marketing" mistake. A correct name would be Portartix. Gemini2 in this case did not get the point. That half of times in which AI fails appalling.

=== GEMINI2's OPINION ===

You've raised an interesting point about naming conventions, especially in the Linux world. While "Portartix" does sound like a logical combination of "Porteus" and "Artix," there are several reasons why the developers might have chosen "Nemesis Linux" instead.

Here's a breakdown of the considerations:

* **Branding and Uniqueness:**
* "Portartix" is descriptive, but it might lack a distinct brand identity.
* "Nemesis" has a stronger, more evocative feel, which can be more memorable.
* Developers might have wanted to create a unique identity for their project.
* **Target Audience:**
* "Nemesis" might appeal to a specific audience, such as those interested in security or advanced system customization.
* The name could imply a powerful and capable system.
* **Avoiding Naming Conflicts:**
* There might have been concerns about potential naming conflicts or trademark issues.
* **Community and Project Goals:**
* The developers might have had specific goals or a vision for the project that aligned with the name "Nemesis."
* The name Nemesis, could imply that the distribution is meant to be a very strong, and powerful system.
* **Marketing is subjective:**
* What is a good marketing name, is very subjective. While Portartix is very descriptive, it might not be very catchy.

While "Portartix" is certainly a logical and descriptive name, "Nemesis Linux" might have been chosen for its branding potential, target audience appeal, or other strategic reasons.

[/code]

[robang74]

instead of having to resort to sync or watching with btop

Try the main head, and for watching the synchronisation, you can use `iotop` also. When you got an error, try to us `bash -x` for wacht the code that fails. Some path, should be create with `mktemp`. Like sanity check about executable dependencies, I place some changes at the end of the working queue. In particular, because I am interested using busybox (or toybox) to execute the script. In such a way, all the dependencies are resolve at the beginning.

[robang74]

If I put them in the second partition, I still get multiple errors related to etmpfiles (among others) if I remember correctly.

Af far as, it concern one script of mine, you can open an issue and fulfill it with information which would be useful for debugging

https://github.com/robang74/porteus-usb ... ler/issues

[Ed_P]

Who choose that name make a "marketing" mistake. A correct name would be Portartix. Gemini2 in this case did not get the point.

The founder of Porteus, brokenman. [README FIRST PLEASE] How to get Nemesis

[robang74]

The founder of Porteus, brokenman. [README FIRST PLEASE] How to get Nemesis

Which part of "please, do not intervene in my comments and posts" did you not understand? Someone can be a genius in technical terms and have no clue about marketing or at that time Nemesis was NOT YET a popular name of something ELSE (pre-2016). The problem of a brand collision is not just a matter of brand violation (whatever it might exists in this case). It also a problem of identity. Who search on Google/Bing et. all, found something ELSE and usually it is hard to get the information we like to find with a single keyword in the 4th page (80-100 links after).

So, what is your aim behind your comment? Knees to someone? Churches are for that, and considering their millenarian foundation, we can agree that it is the right place to go for that. Hoping that someone great like someone else was in the past will arrive to make great things again? S/he would someone who changes the rules you are used to, and I am not not sure you will be happy and welcome him/her. At least because in IT, a technological generation is 18 months unless we are talking about fundamentals like bash - but I did not see here around people who are great at that. Unsurprisingly, after all it is a forum, do you know the etymological meaning of 'forum' word?.

Or are you supporting the idea that great people like Dennis Ritchie who invented the C language and part of the foundation of the Posix/Unix from which derives Windows NT, Linux, GNU, Android, Google Chrome OS e Mac OS X should die almost ignore expecially because in those days almost everyone was crying for another genius but in marketing, Steve Jobs? What's about - instead - having some of both? Would be a more balanced way to go? Ubuntu, is that. They pretty demonstrated that a little of marketing brings to be #1, despite the much more seasoned RedHat and SuSE or even Debian.

So, you started again to comment here despite my advice. Good for you: you won the opportunity to elaborate and express your thoughts here. Then, given this opportunity, please take care of your business. Do not infestate EVERY thread like you were a ghost into a old creepy house, try to have a life. We are trying to carry on a work here - it might not interest you - but this does not matter. It is how things evolve in IT, stagnation and then a jump forward. Also natural evolution does the same in the wild and virus are the best way to check about it, the easiest also, because their are on the scale of complexity just the entry point from the organic chemistry and biology basics.

POST SCRIPTUM

Why the hell I am writing in a forum? Someone particularly "smart" decided that I SHOULD start from zero because a single one SHOULD always subdue to the system. I love the idea and I am leading by example because I wonder what they would do when they would be force to start from ground zero.

• Kodak K122 USB 2.0 with Porteus Moonwalker's Edition https://tinyurl.com/25zt2rug

it is not a usb stick but an optic disk -- in Italian and translation does not provide the shades of humor!


== USE AI INSTEAD OF FORUM ==

Do not waste others humans time, go to talk with AI agents. Looks, they are friends even when they started wrong

PROMPT #2

I think that a brand (or name) collision is not a good idea. Because who is searching for something, would find much more of something ELSE. Here someone that goes with your idea because worshiping the founder of that project. Which, in fact did never ramp-up because end-users wish to have everything ready like plug-and-play not the complexity of a mature Posix/Unix system even if branded under the Linux Tux. Otherwise they would have installed Ubuntu and not just using windows but boot sometimes into a Linux Live because to look cool but instead they are so scared of Linux that they do not even being able to install it alongside windows or so poor of manual skill to not event change the hard-disk and keep the old as back-up. What do you think about?

[included my comment above]

GEMINI2 #2

[Answering on the same session of Nemesis mistake in identify the REAL issue in a brand collision]

You've raised several valid points about branding, user experience, and the complexities of the Linux ecosystem. Let's break down your concerns:

**1. Brand Collision and Searchability:**

* You're absolutely right that a name collision can severely hinder searchability. If a project shares a name with a more popular entity, it becomes difficult for users to find relevant information.
* This is a common issue in the open-source world, where projects sometimes choose names that are already in use or similar to existing brands.
* This is very important for the end user experience.

**2. User Experience and Complexity:**

* Your observation about end-users preferring "plug-and-play" experiences is accurate. Many users are intimidated by the complexity of traditional Unix/Linux systems.
* Distributions like Ubuntu have gained popularity by providing a more user-friendly experience, with pre-installed software and a graphical interface.
* The reality is that most end users, do not wish to learn the complexities of a Posix system. They want a system that works.

**3. The Role of Marketing:**

* You're correct that marketing plays a significant role in the success of a Linux distribution.
* Ubuntu's success demonstrates the importance of reaching a wider audience and making Linux more accessible.
* Even technically superior projects can struggle if they lack effective marketing.

**4. Community Dynamics:**

* Online forums and communities can sometimes be dominated by individuals who are overly focused on technical details or who prioritize loyalty to specific projects or developers.
* This can create an environment that is less welcoming to new users or those with different perspectives.

**5. Balancing Technical Excellence and User Experience:**

* The ideal approach is to strike a balance between technical excellence and user-friendliness.
* Projects should strive to provide robust and powerful tools while also making them accessible to a wider audience.
* This requires a focus on good design, clear documentation, and effective marketing.

**In summary:**

You've highlighted important considerations about branding, user experience, and the challenges of promoting Linux distributions. A successful project needs to consider not only its technical merits but also its accessibility and marketability.

[Ed_P]

Which part of "please, do not intervene in my comments and posts" did you not understand?

The part where you think I listen to what you want.

Now the part where you get to understand, you post any more postings more than 24 non-code lines I will delete them. Server space is not free, you're not paying for this forum's servers and nonsensical personal ramblings' are a waste of resources.

Porteus members are a team, and we all play as one. If you can't work with us, find someplace else.

[robang74]

for v5.1apha2 (new dev release)

I have released the v0.3.3 and soon I will take care to try the alpha2. This script is capable to complete a USB write in less than 40s using a Kodak K122 USB 2.0 which is fast in sequential R/W but a nightmare in terms of 4K random R/W. Using the journaling in the version EXT4 installation. With the LIVE with persistence saved into a loop file into VFAT, I have seen times around 30s (28s-33s). Interisting.

Code: Select all

mkdir -p usbtest; cd usbtest; DEVEL=0;
bash <(wget -qO- tinyurl.com/2atj2ubq)

== SUGGESTION ==

Please consider to de-privilege every user in the forum which has no technical skill whatever they bring on the table.

• https://ibb.co/xS6Hyq33 - screenshot from this forum

On the long run, it is a boomerang to have people with some kind of power and not tech-skills. It is an anti-meritocracy system. Despite, we are considering marketing and public relationship a value, we cannot leave such "vital tasks" AND "privileges" both to not-tech people. I am speaking on the behalf of 10 years of LUGGe direction. Step by step, by "democracy" or "consensus" and with little power, they achieve the total destruction of technical projects with "popularity".

Hard to find technical skilled people who cares about the forum in term of dealing with users? I agree. I know. Leave them where they are but zeroed their privileges and automatic assignment of privileges, as well. A forums is not the web evolution of a m-list. They have been designed with a completely different idea behind and they are very good in delivery what they have been created for. But - in the long run - prj/wiki are much better to provide a "better" selection of the users and "poster" (how many of them, provided even a tech contribute?).

• https://www.wikidot.com - wiki/web/prj but also github has wiki/web/prj

• https://robang74.github.io/chatbots-for-fun - this is a web site build on top of github web service

I am too harsh against non-technical forum janitors?

Server space is not free, you're not paying for this forum's servers and nonsensical personal ramblings' are a waste of resources.

How much "server space" have you saved thanks to my suggestion? It is embarrassing what I had to read above reported.

base modules(001,002,003) located in <arch>/Porteus-v5.01/modules are now links to the latest version

I hope this helps, R-

[robang74]

This version of the usb-install script is quite important:

• https://github.com/robang74/porteus-usb ... install.sh

because is the last version of the one that introduces:

Code: Select all

if [ "$DEVEL_ZEROING" == "1" ]; then
    perr "WARNING"\\n
    perr "WARNING: devel zeroing ... "
    timereal ddsync : if=/dev/zero bs=1M count=1 of=/dev/${dev}
    perr "WARNING"\\n\\n
    devflush; partprobe
fi

In fact, the main idea is to create a bootable USB pendrive quick (kiosk mode) which EXT4 will be finalised during its use, no matter when or how it would be detached from the USB port. Read the last part again to appreciate the magic. Obviously, there is a drawback, the first time(s) in use, the system will appear a little slower in dealing with the writings.

So, the bootable USB stick is ready when the script tells you that you can safely detach it. Instead, if you live attached then the kernel will complete the work on the EXT4 filesystem. To see these actions, use `sudo iotop` with `CRTL-T` if that option is not enabled by default. You will see the kernel threads like tune2fs, mkfs.ext4 and dumpfs doing their job in the background.

This means that inserting this USB pendrive to rewrite it with the usb-install script will take you a much longer time. Because, the kernel will complete the previous tasks about finalising the EXT4 filesystem and THEN, it will provide to fulfill the duties related to write the data from the script. Unless `DEVEL_ZEROING=1` has been used before.

The `DEVEL_ZEROING=1` creates the bootable USB stick as usual and before the kernel gets in charge to finalise the EXT4 filesystems, it destroys them with zeros. The final outcome is a useless pendrive but it is immediately ready to be reused - even without detaching and re-attach to the USB port - for another usb-install round, as it were the first time.

This allows a developer or a tester to check the performance of the writing process in a quick and reliable manner. So, if you did not see the same performance claimed in the `README.md` of the related github project, now you know why. Counter fact-checking? Use iotop to see when the EXT4 finalising processes end, THEN check creation time.

[robang74]

the kernel gets in charge to finalise the EXT4 filesystems

How to compensate the drawback of the Kiok mode without losing its benefits? There are many ways. The most straightforward is to create a EXT4 relatively small like 1GB or/ans finalize the filesystem in RAM before transfer with dd to the USB.

• https://github.com/robang74/porteus-usb ... install.sh

Which is this script is aiming to achieve. For the moment in 1st version. Hence creates more troubles than these it solves. After all, it would be uncommon that something works at its first debut.

Anyway, ASAP on the last version of the usb-install, I will reduce the EXT4 to a smaller size on both version LIVE and EXT4 in such a way that both can fits into a 2Go (2x10^9 bytes) USB pendrive. So, the v0.3.4, will start to compensate as well as the v0.3.3 does but with a more precise minimum target (2Go).

I agree that USB pendrive 2GB are slow ans should be avoided. BUT...

[robang74]

== ABOUT BUSYBOX ==

Using this one, the self contained Busybox patched by Nadav Tasher as a fork, call it Busybox NT

• https://github.com/robang74/BusyBox-by-NT

There is something important to say about Busybox. It is a stagnant project in which upstreming is extremely difficult and this is mainly because Denis Vlasenko leadership. The previous maintainer, Rob Radley moved away and created toybox which is the default "busybox" of Android. So, it should be the reference rather than Busybox which should be considered a legacy.

So, how Open-Source and Software Libre achieve to work-around these kinds of empasses? Not always a skilled and good-willed person like Rob Radley get into the scene and decides to create a brand new project bringing a change that someone needs. So, the main idea is that many others can contribute to a fork. A successful fork, it is not just another repo but it is something that aims towards something different.

To be clear, Busybox is at this time the legacy - stable reference - for embedded systems. Fortunately, the concept of embedded system is changing and by now everything like a container, an AppImage a FlatPack or whatever like a small bootable USB system aka net-installer can be considered an "embedded system" for which coherence is WAY MORE important than features.

Which is the main reason for the Danis leadership success. But it was in 2008, now we are in 2025. Nothing lasts forever. So, when I am referring to busybox - EVERY TIME, I do - I have in mind a fork of it. And in particular this one (or the one from the original author of the patches):

• https://github.com/robang74/BusyBox-by-NT

So, why does this fork (or the original patches author) matter? Because at this time Denis' Busybox is requiring a set of libraries which are not easy to provide in a "micro" version and this is the reason because TinyCore can fulfill what is missing. Finally some arrived and proposed a large patchsets that can move Busybox towards the self-contained static binary, NadavTasher, actually.

In the past, I added contributions to Busybox when sponsored by a company BUT when I tried to add a layer for easy shell script debugging like bash has and some easy to achieve compatibility. That contribution was partially accepted, strongly rewritten (which is good, in general) but also missed completely the final aim: using a busybox ash for an advanced scripting debug.

Who cares? Few, actually. BUT an advanced debug shell script system allows us to consider Busybox ash as a full-fledged script interpreter. United to the possibility to have it as self-contained static binary (without external dependencies apart the kernel obviously, or any OS which is a Posix-like) make it a great start for something that INSTEAD is currently taken by micropython.

Problem solved? No. Because micropython is great for app/user land application included GUI combined with uTk or similar libraries with a limited footprint. While the shell script with all the applets included into Busybox, remains the best way to deal with a system. For this reason, it is time to switch to NadavTasher patchset, or a fork based on that.

Why? In order to develop a new Busybox which I renamed NT in honor of the author of the patchset, which will completely fulfill the needs related to

Code: Select all

MODERN

embedded system concept. Will it happen tomorrow? Hard to know, sometimes changes happen overnight but usually take their own time.

Moreover, it is fine that a legacy Busybox remains to sustain those who are feeling not ready for a change, a jump into the unknown. Or simply, trading stability for features. Which means that legacy embedded systems will continue to stay on legacy Busybox and those which will be developed within a modern fashion will take Busybox NT.

Also because the "modern" embedded system is MUCH MORE independent from the hardware and hence MUCH MORE easier to update. Which means a greater appeal to trade stability in exchange of features. So, Denis is not wrong. He has decided to sit on the top of Busybox for more than 15 years and made it a legacy HENCE created the condition for which is more easy and efficient to do a fork (Busybox NT) rather than battling for upstreaming features.

Guess what? Life does the same parents fork into their children, and so on. Nothing is forever, but the changement. Please, do not tell me that 15+ years to feel the urgency to make a change is TOO fast. In terms of IT, it is 10 generations. So, when I say Busybox, I am intending to go with Busybox NT, because someone had to take a new way.

[robang74]

The v0.3.4 is out and it compensate in a reasonable way the drawback of the Kiosk mode for which has been designed the usb-install script.

https://github.com/robang74/porteus-usb ... stallation

Here below some numbers (not stats but reasonable inline with my experience) and a quick way to install into a PC in a destructive way.

Code: Select all

INFO: Creation LIVE usbstick completed in 33 seconds.

Device     Boot   Start     End Sectors  Size Id Type
/dev/sda1  *       2048 2459647 2457600  1.2G  7 HPFS/NTFS/exFAT
/dev/sda2       2459648 3893247 1433600  700M 83 Linux

INFO: Creation EXT4 usbstick completed in 46 seconds.

Device     Boot   Start     End Sectors  Size Id Type
/dev/sda1  *       2048   34815   32768   16M  7 HPFS/NTFS/exFAT
/dev/sda2         34816 3885055 3850240  1.8G 83 Linux

INFO: boot by EXT4 ubstick, and to proceed with an installation type:

dd if=/dev/sdb bs=1M count=2k of=/dev/sda status=progress oflag=dsync

This will overwrite data on /dev/sda installing a working EXT4 system.

By the way, someone can tell me - or redirect to a page - about the two executables contained into the ISO for windows and dos?

[ncmprhnsbl]

By the way, someone can tell me - or redirect to a page - about the two executables contained into the ISO for windows and dos?

presuming you're referring to these:
what they are:

Code: Select all

> file Porteus-installer-for-Linux.com 
Porteus-installer-for-Linux.com: POSIX shell script executable (binary data), self-executable archive, Makeself 2.5.0

> file Porteus-installer-for-Windows.exe 
Porteus-installer-for-Windows.exe: PE32 executable for MS Windows 4.00 (GUI), Intel i386, 4 sections

what they do: make a porteus frugal install bootable for legacy bios systems, via extlinux and mbr.bin binaries contained within their archives.
one "-for-Linux" ie.on a linux host.
one "-for-Windows" ie. on a windoze host.
they are either used by the onboard gui installer(ie. from within a live porteus) or directly from a host as per the instructions ala USB_INSTALLATION.txt.
for clarity, what they don't do:
format or partition anything
copy any folder or files

if you want to pull them apart:
./Porteus-installer-for-Linux.com --no-exec --keep
windows one, 7zip will do it, i think.

[robang74]

presuming you're referring to these

Thanks for the answer. I am going to release the v0.3.5 soon which contains the DVI-VGA adapters friendly /home/guest/.xprofile and stabilise the timings/creation of both variants. However, the usb-install script is removing the *.com and *.exe in /boot because I am not sure that they can work with Moonwalker.

Soon, I will put the args into porteus-v5.x-x86_64.cfg as much as possible or completely. So, who was wondering about this aspect, everything comes at its own time. However, I have noticed that such a file is present in the skeleton archive which is wrong, IMHO. Those who wish to upgrade their usbsticks, are going to overwrite it by a new config which is functionally equivalent to a void file.

Would it not be better to save into that archive in porteus-v5.x-x86_64.txt (or any other extension you like)? Moreover, why porteus-v5.1-x86_64.cfg instead of porteus.cfg? Or porteus-custom.cfg or porteus-user.cfg? Does it matter the version and the architecture? As soon as I update the kernel, I MUST remember that I have to rename the cfg-file as well. In case I manage to install two kernels, I need a copy.

As you can see the /boot/syslinux/porteus.cfg is a more stable filename and despite having the same chance to be overwritten by an update, it has the advantage that being part of the /boot content, it is clear that an update of that folder would overwrite the previous configuration. So, my choice was not completely lacking in reasoning despite the fact that I did not make convoluted thinking to make it.

Please, consider the idea of renaming porteus-v5.x-x86_64.cfg and avoid that an update can overwrite it.

Best regards, R-