Deepin prepares to leave Debian base and move to fully independent distro A leading Chinese Linux vendor is polishing what may be its last Debian-based release, and preparing for the move to becoming a fully independent distro with its own new package format, Linglong. Deepin, the most internationally visible Chinese Linux distribution, has put out two new versions: the latest release of its …
I was really excited that if China, with all its manpower, decided to adopt Linux desktop as a mainstream, they would commit upstream to a global distribution like Debian. These increased resources would make Debian improve a lot quicker and maybe this would make is the OS of choice for people who didn't want to pay for Windows or Apple. Sort of how like China investing in Riscv is leading to it becoming the CPU design of choice for the future.
Instead they fell into the NIH syndrome and linux distros remain heavily fragmented. :(
Mandarin isn't a script, you know. It's a spoken language, and you can write it in ASCII characters. Since C is written in ASCII, Chinese C has to be, too.
There is an entire ethnically-Han Chinese-speaking community in Kazakhstan and other former-Soviet territories who write their language in Cyrillic.
https://en.wikipedia.org/wiki/Dungan_people
C source code written in Chinese by a Chinese speaker is still C.
It might say
for (zhishu = 1; zhishu < 11; ++zhishu)
{
printf("%d ", zhishu);
}
return 0;
I don't think anyone would have a lot of trouble working out what it did.
>Mandarin isn't a script, you know. It's a spoken language
My bad, for not getting the right script language.
My thoughts weren't so much around the actual C code - although it is a relatively trival change to the compiler to change key words to alternative character sequences, but the comments describing the code and that the module would be buried on a Chinese version of GitHub on a Chinese website; so not unlikely to be locatable via a standard Google search by a typical US/Europe-based non-Chinese developer. Plus I suspect Richard Stallman never thought about such matters and simply assumed programming was only done in English.
Trouble is having to translate stuff is a big hurdle...
A while back I was having to do some work that required the referencing of technical drawings. Whilst I could read the drawing (the object lines) and make intelligent guesses about dimensions as to the units were in Arabic numerals, the annotations were problematic due to them being in sinograms.
AppImage has some deep flaws. It maintains a hardcoded list of libraries which are not included in the AppImage. The result is very often a crashing application, because the host does not provide the library or the host version does not match the linked version. AppImages also crash if FUSE is not available. You can tell AppImage to self-extract and avoid FUSE, but it requires an environment variable. Why not just fall back to self extraction instead of crashing? Goddammit.
Flatpak fares far better. Looking at the LingLong docs, it looks very similar to Flatpak API. Maybe it creates the needed competition for Red Hat to polish Flatpaks rough edges.
Snap meanwhile ... is just a half-arsed Flatpak competitor and ought to be scrapped. Snap out of it, Ubuntu!
The article briefly mentions Flatpaks focus on desktop. Would be interesting to see LingLong on the server. It seems to rely on Linux containers for its sandbox, so could certainly tackle Docker (if LL would export OCI container like Flatpak does LL would even be compatible with Docker). I'd also welcome an effort to replace Kubernetes with a revised, less complex component, based on LL capabilities.
From your description, it sounds rather like FlatPak, Snap etc are equivalents of Windows installs where all the relevant libraries etc are provided with each app, bloating them up. Am I on the right lines here?
Full disclosure. I use FreeBSD primarily, with Linux and Windows coming a poor joint 2nd place :-)
The closest thing to a Windows .exe file is AppImage. AppImage builder puts your ELF binary and libraries (except those from the exclusion list) into a file system, which gets baked into a big ELF file. Yes, there's very much the same duplication issue, as you suspected. Also no sandbox to help contain the exploit.
Flatpak tries to do better. There are runtimes, archives of common libraries, including the Freedesktop stack. One might say they created the one and true Linux distribution floating above the host distribution, relegating the host distro to a diminished role. Though quite a few developers want it and users don't mind (there are shiny icons in the app store). Still, many apps bundle their libraries and Flatpak devs acknowledge there's a slew of outdated or vulnerable bundled libraries out there. IIRC they are setting up a scanning service.
-> There are runtimes, archives of common libraries,...
To me this is an indication of failure somewhere along the line. It is with reluctance but also objectivity that I point the fingers at several things:
1. The lack of a "standard" Linux distribution, right down to the shared library level. I did my time searching for rpms with exact version numbers otherwise certain apps would not run. rpmfind was a saviour years ago, and it's still around today. This is the equivalent of "dll hell" in Windows land, so it raises my hackles when I see Linux proponents ignoring that it is alive and well in Linux too.
2. The package managers (all n of them) seem unable to handle all the dependencies from point 1.
3. The number of package managers is a nuisance. If you spend time writing and building a Linux app, you can either:
3a. Provide it in source form, which is not the most user friendly get-up-and-go way of doing things. If i just want to try an app I don't necessarily want to build it, I just want to use it. This is different from providing the source code for the app, which >99% of users will never look at.
3b. Package it up n ways for all the different package managers, and maintain that ongoing. This is duplication of effort no matter how you look at it. However, as mentioned in point 1, there is still the issue of the lack of a standard Linux installation.
3c. Provide snaps/flatpacks/linglongs or whatever else comes next. Each of these has it own drawbacks, as mentioned in the article. I give Liam Proven a nod here for not just glossing over the drawbacks, as though they don't exist. Also a nod to Duke of Source for mentioning the bundled up libraries which become vulnerable over time - these are now outside the normal update loop for the main OS. Your OS may be up to date, but your apps are not.
I know that the Linux community likes to say that choice is a good thing, and in many cases it is. But sometimes choice just causes a lot of duplicated and wasted effort.
[Article author here]
Deepin is the international version of UOS from Tongxin UnionTech.
And the desktop is not KDE. DDE is its own thing and it doesn't work like KDE in any way. As far as I can tell, you can install both UKUI and KDE on the system at once and switch between them.
My personal 2¢ worth: I prefer it to KDE. It looks better and for me it works better, but it is a lot less customisable.
I have only seen the screenshots of UKUI, but my immediate reaction was, it sure looks nicely polished and user friendly. It's good to see competitors emerging and challenging the two big desktops (and their offspring). System76 is another horse in the race with their announcement not too long ago (https://www.theregister.com/2021/11/08/system76_developing_new_linux_desktop/).
The CCP require certain apps be installed on all platforms.
I suspect that Deepin will have these pre-installed. No matter how lovely their spin is, it is what it is - CCP controlled under the bonnet.
It's rather sad that we can't use and really appreciate the hard work that has obviously gone into this distro.
Cheers
Jon
As I said in the piece: there is also an Ubuntu remix based on the DDE desktop.
If your distrust of Chinese software is so great that you won't use a Chinese distro, then I suggest that you go and help the Ubuntu DDE maintainers to update UbuntuDDE to 22.04.1.
If that is not your area of expertise, then perhaps run Deepin in a VM and monitor its network communications and see if it phones home. I think a lot of people would be interested in the results of that.
Me personally, I occasionally use the Opera browser (now Chinese-owned) and my one remaining Windows 7 machine runs Total360 antivirus (also Chinese). My smartphone is Chinese, bought from China directly, and I strongly suspect that my other smartphones were made there, too.
They all seem like good products to me, and I am not as paranoid about Chinese software as you appear to be.
But if you can collect some evidence that your suspicions are founded in reality, then that would be very useful and a valuable service.
Quote: ".....an independent upstream......."
.....but how "independent"? Will there be a Chinese replacement for gcc or llvm......or will the new "upstream compilers" be simple(?) forks of the existing tools?
Not an expert here, but this plan sounds like a lot of heavy lifting! And for what benefit?
The full quote puts things into perspective.
"to provide basic services and the foundation for the establishment of an independent upstream."
So they have done the "simple" step and created their own distribution. This permits enlargement of the ecosystem in two directions, firstly it gives a target platform for downstream (Chinese) applications development, secondly, it gives a target from they can work backwards from and instead of taking (non-Chinese) upstream packages they can fork these packages and establish their own (ie. Chinese) expertise independent of the USA and its allies...
Given the size of China with respect to the USA and their ethos, I expect they will pull it off and some! Would not be surprised if within a few decades the maintainers of say gcc and llvm will start to take updates from China...
Yes a lot of heavy lifting, but if you are playing the long game, my bets are on China winning; unless the US, UK et al.go through a mindset change...
@Roland6
Agree!!! China has been in "the long game" since 1980. I recommend an easy read, which helped me understand where their "long game" started:
- "Deng Xiaoping: The Man who Made Modern China", Michael Dillon
For reference, the definition and execution of the long term plan created by Deng can be seen in two GDP facts (GDP 1995 to the present):
- UK GDP has doubled
- China GDP has multiplied by 17!!
No......this AC is not a communist.....but this AC is definitely a fan of long term plans. Clearly we need one in the UK!!!
For certain kinds of decision, however, China is *tiny*, having only a handful of people who are allowed to think.
You may like to believe that economic and technical questions don't fall into that category, but just ask yourself how many times politicians in the west interfere in such decisions. It's a nuisance here. It's the end of the discussion there.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
