Are flatpaks better, worse or the same as snaps for taking up space?
It seems trying to fix the 'terrible binary compatibility across the Linux/open source ecosystem', has only caused more fragmentation.
Are flatpaks better, worse or the same as snaps for taking up space?
It seems trying to fix the 'terrible binary compatibility across the Linux/open source ecosystem', has only caused more fragmentation.
Well regarding raw space I think flatpacks are not as large as snap but there is a "but" - it's not as large of a difference when you take into account that snaps installs all dependencies in itself. Yet again a contra-but argument - snaps always saves at least two versions after updating (newest and the last before update version in case of an error so you don't get stuck and stuff still can be retraced easily) - and that takes space as well but on the other hand it grants a bit more security
The way I see snap vs flatpack is security/stability vs speed/space because snap in my experience really gives more polished but larger apps for last year-and a half
But in the end - internet forums are your best friends because it gets to the point of playing coin-toss: "Will snap app x be better or will flatpack version of the same work best for me?" - and usually there is no clear winner because sometimes flatpack works better and sometimes snap gets the job done faster (less feral hunts for missing dependencies at a cost of few MB space for example)
Also - there is a third option I tend to bump into from time to time - Appimage but I don't have much experience with them so idk what to think about that
There is also FOSS question at hand but it gets more into ethics than into practical side of things so I won't dive into it here
In the end snap package will most probably be larger than flatpack equivalent (as of writing this post haha)
I can't seem to paste a table here, and manually entering table data is a bit of a burden, so let me just capsulize the results of a ChatGPT investigation I performed against four different distros I'm considering:
Final Thoughts
Voyager: Great if you love visual customization and a unique UI.
Linux Lite: Perfect for older computers and users switching from Windows.
Linux Mint: The most polished and beginner-friendly option, especially with Cinnamon.
Ubuntu MATE: A lightweight, traditional Linux experience with modern Ubuntu support.If you’re new to Linux, Mint is a safe, well-rounded choice. If you need a fast system, Lite or Ubuntu MATE are better. And if you love unique visuals, Voyager is the way to go!
I like Ubuntu MATE's interface over Linux Lite (which is too Windows-like for me). Voyager is a delicious interface, so I think I may give it a road test. I'll switch over to Ubuntu MATE if I find performance too slow on my Yoga.
Does anyone know why the distro install doesn't have a step presenting a checklist of "optional" packages (packages designated distro-core bundle, but not necessary for functional system, i.e. Firefox, Evolution) and allow the person doing the install to un-check those he wants to skip, before moving on to the actual install commit phase ?
Canonical built Ubuntu's installer subiquty, I would guess it was a UX decision to keep the "standard" and "minimal" option that the older ubiquity already had. Under the hood it just installs one of the two metapackages ubuntu-mate-core
or ubuntu-mate-desktop
. Only their developers (or their Discourse) may know the real reason, or a bit of history when it was first added.
I know EndeavourOS' installer (Calamares) gives you a choice as you describe — down to the kernel, desktop, applications, optional drivers, etc. EndeavourOS is a good one for anyone wanting to a smoother sail into the Arch & rolling release world.
Fun fact! Our project leader was once an Arch user and maintained
mate-desktop almost a decade ago!
I see the Linux desktop as [distro base] + [desktop environment], so I don't have any desire to experiment. It would only be for development purposes if my users have very distro-specific bugs.
No one's really mentioned it yet, but if one loves MATE but not Ubuntu or corporate backed distros as much, there's all the different distro bases - particularly community ones like:
+
=
+
| Arch Linux + MATE
+
| Debian + MATE
+
| Fedora + MATE
I tried and liked EndeavorOS, but didn't want to shift the whole core/kernel thing. It's funny how we can get locked into a paradigm in a field where choices abound. Even though I like playing with / sandboxing systems, I always seem to come back to / stick with Ubuntu as the core.
The interesting alternate might be immutable distros. Unlike traditional/classical distros, the root file system doesn't change between reboots. Great for security and consistency, with personal files / programs (e.g. as Flatpaks) kept separate. Not great for flexibility and full control.
Reminds me of Chrome OS and non-rooted Android devices.
Examples: SteamOS, Fedora Silverblue and Ubuntu Core.
Indeed immutable distros sound like a nice "near future" thing that might just help propel linux as a "fool-proof" desktop environment for people that really don't need extra layers of flexibility and love to keep things just the way they are and up and running
Only barrier for now it seems to be installing things like Lutris and other useful apps since it looks a bit too complicated of a hustle compared to simple apt terminal commands but that just might be me being used to the "standard" linux way
One main thing I love about the immutable distro idea is that it could narrow down problems and help find solutions since everybody would know what others have added since the base would be standardized for every user - I like to think of an immutable distro as a "capability constraint" (an idea from structuration theory by A. Giddens)
For example even though it constrains the user from fully being free to do what they want it also enables the user by giving them a firm ground to build, it provides a framework to build around and develop while knowing that if more users have the same exact base framework the solutions for potential problems will be easier to find because what appears to be a constraint may actually be a safeguard - maybe that would be a way to a more cohesive, unity focused foundation for all distributions since it seems that linux fragments its community too much to be taken seriously sometimes
Well, along with the immutable distro idea, I think there are a couple of things that are preventing a wider acceptance of Linux. Here are a few (I'll probably think of more as this discussion continues...)
Kid3-qt
, do you have an idea what it is or does (it's an editor for tags in MP3 files). How about Pluma
? I like it as a text editor, but the name doesn't evoke images of pens, pads, note-taking, or similar.The Linux community has created a self-licking ice cream cone. Everyone is free to contribute, fork, and share their work. But that lack of consistency is self-defeating.
"Get a Tuxedo!" - No need, already got one in the wardrobe, thanks! (Couldn't resist a pun )
Your point about application names made me think of elementary OS - all their apps are tailored made for their design language and named after what they are. I'd say they're a distro taking a more "opinionated" approach (like Apple) where it can feel all cohesive. It's not perfect, since upgrading major versions requires a full reinstall.
Renaming application entries is easy - but if all the file managers were called "File Manager" in Ubuntu, who's who? "Broken system showing me two of the same!!" exclaims the user ... well.... could be worse, there could be two control panels!
Say, when it comes to toolkits, I noticed that KDE put in the effort to make GTK apps (e.g. GNOME Disks) look part of the system (Breeze theme). Can't say that about Ubuntu since Qt apps look nothing like Yaru on GNOME - but at least
Ubuntu MATE ship the necessary configuration (... at least, it use to & did under Ambiant-MATE)
Makes me wonder, will it be a non-profit, a corporate entity or a community distro that brought the general Linux desktop into mainstream? Will it be flexibility (KDE/MATE) or opinionated workflows (GNOME) that prevails?
Other thoughts:
So, Luke, is the entire root partition locked out for the "immutable" scenario, or only those parts which make up the OS itself?
Would user-installed Apps drop
And just how "immutable" are we talking about?
completely frozen?
modifiable ... but via an "overbearing" (acceptably so) process intended to protect the system from all but the truly knowledgeable and experienced ?
So, why use caja
for the file manager? Even its meaning in Spanish doesn't fit: Box, crate, cash register, etc. And even if it made sense in Spanish, why use it in an English deployment? And pluma
is Spanish for feather, quill, pen, etc. I don't speak Spanish, so how am I supposed to know that pluma
is a text editor. Windows has notepad
and Apple had TextEdit
. Those names are both English (the locale
of my installation, and are quite descriptive. I don't have to guess at what they are or what they do. This is design work, and poor design can break the best of products or ideas.
I'm holding in my hand a book I purchased in 1992, Human Interface Guidelines: The Apple Desktop Interface. © 1986, Apple Computer. $14.95 (about $57.26 today, for a book slightly thicker than a graphic novel). I love that book! Because it's not a computer book -- it's a psychology book! It's how people interact with computers and their visual elements (menus, dialog boxes, radio buttons, check boxes, scroll bars, etc.). In the early days of the Macintosh, if developers didn't conform to the consistency of the interface, their products didn't find acceptance in the market. That Linux still has a percentage of the market is a testament to those of us with an adventurous spirit and a high tolerance for pain!
Well, Fred, I will concede my own personal preference for all things technical and computing to be in English, and this coming from a very proud French-Canadian born, bred, and surprisingly still speaking that very poetic ... and diplomatic ... language!
However, to suggest that English-specific terminology be exclusively used for the actual naming of applications running on English-centric hosts really, seems a rather onerous imposition on developers, because it seems that you are suggesting that these developers should, per what you said, come up with a similar-meaning word for the primary (and secondary? or tertiary?) language of each of the 195 countries in the world, or would you be suggesting it should be extended for all who wish it among the 7,139 languages on this planet? ( Yes, I am exaggerating to make my point!)
Notwithstanding the workload of such, you seem to imply that application creators cannot choose the name by which they want the legacy of their creations to be remembered by!
If you stop and think about it ...
Is that really what you want to say?
Do you really want to take that away from all those generous coders sharing their results of all their hard work, only to be remembered by a label that they don't really want associated with their product?
I couldn't imagine myself even thinking of suggesting such a thing to creators.
Applications are published, referenced, discussed, supported across various forums by referencing the tasks being performed by those applications, along with the chosen, unique label, which is necessary to facilitate the growing of, and recognition of, the associated mind-share as the application's popularity grows. Multi-naming would likely dilute that to the point of preventing applications, such as Gimp, from emerging as segment leaders.
No Eric, I'm not suggesting that at all (well, not seriously). What I am saying is that I think many people shy away from Linux because of the odd names applications and processes they see. Those of us who aren't afraid to roll up our sleeves and get a little dirt under our fingernails think nothing of opening a Terminal session and running top
, or ps
or a limitless number of commands. Or writing a script. Or even choosing a different terminal emulator! But many people don't want anything to do with a command line.
True story: A number of years ago I was in a meeting with a customer and was asked, "Does your product have an API into your logging system?" I answered that no, we didn't, but they could simply tee
the log to whereever they wanted. Or they could use a named pipe
if that would better suit their purpose. The reply they gave me was, "But we want a GUI." Head scratcher. They already had the capability at hand, but were asking us to build a GUI into log writing!
When I install Ubuntu Mate for others I just rename the desktop Icons but not the menu items like Pluma Text Editor, Synaptic Package Manger and LIbreOffice Writer. That way they know the name and purpose.
It's nice of you to do so (and you've made my point for me), but most people don't have the benefit of someone to install and configure their system for them.
Definitely a head scratcher, that one!
A post was merged into an existing topic: And For My Next Trick
You definitely have a point,but on the other hand: Windows is equally weird with their app naming:
So in the end it all comes down to familiarity and we already lost that battle.
( except for GNOME which does what you propose:
Gedit: editor
Files: filemanager
FileRoller: file/dir-compressor
etc. )
You could have just stopped at
Windows is equally weird