France starts moving government systems from Windows to Linux

[RoGeorge]

Well, either Windows or Linux are computer related, except the French don't use computers.
Instead of computers, they have ordinateurs. 

[Cyclotron]

You just used the term "value proposition" when addressing a nation of people and then claimed they are all in it for the sake of greed. That's ironic.

What's ironic is that the word "nation" comes from Latin "born", which is the opposite of a population whose only shared trait is having sought better life by immigrating to some place at some time. Whole concept is a greed magnet, attracting the greediest of the world as surely as socialism attracts the laziest. And the value proposition of this social construct has been deteriorating for a while, including before the particular administration which some love to blame for all ills.

The US citizenship is about 85-90% born in the US vs. naturalization. In Europe, the number is about 90% of the citizens born there. Not a huge difference, is it?
Now, Poland, the flag you fly; that country can brag about 98% of its citizens being born there, which is about a match for Japan.

Based on your argument, would you say there are no nations? Would you claim that the entire world was populated by immigrants at some point, and therefore that there are no nations because the word "nation" has a Latin source of "born"? All of them many times over too by different tribes and different sources of humans.  This argument really doesn't make any sense, does it? Socialism seems to have attracted its fair share of greedy people, hasn't it?

https://www.forbes.com/sites/sylvanlebrun/2025/04/01/the-countries-with-the-most-billionaires-2025/
Note that China is actually higher since they count Hong Kong separately for some reason.

For more recent numbers, see here:
https://en.wikipedia.org/wiki/List_of_countries_by_number_of_billionaires

If you go by the Huron list, the US is second to China.

Greed for money and power is universal. The US certainly has problems, and hence the claim to be working towards "a more perfect union" is a goal. It's only been 250 years this year, you know. Some roads and buildings in other countries are older.

[Cyclotron]

I see this movement by the French government as something only governments can afford to do. Maybe startups, too, since they don't have a lot invested.

Form where I'm standing the world has moved away from MS-Office. I deal with quite a few small companies and startups. The past 10 years the requests to use MS-Office document formats have dried up completely for me. I see most customers using Google docs or LibreOffice.

I think that agrees well with my views. As I mentioned, startups rarely invest in learned behaviors and workflows, and while many can bring that baggage with them from other companies, startups are often young in their staffing.
This is exactly why Microsoft, Cisco, etc., spend tons of money each year on the EDUs. Getting folks in the ecosystem early pays off later. Google does this too and, by and large, is getting its fair share of young people exiting university with a penchant for Google software and email.
It also happens to be the source for the most successful Linux distribution (Android).

[free_electron]

I see this movement by the French government as something only governments can afford to do. Maybe startups, too, since they don't have a lot invested.

Form where I'm standing the world has moved away from MS-Office. I deal with quite a few small companies and startups. The past 10 years the requests to use MS-Office document formats have dried up completely for me. I see most customers using Google docs or LibreOffice.

Google docs is pretty good ,so are the online versions of office. Presentations, documents (powerpoint / word) not really an issue. The sticky point is excel ... None of the alternates can run excel macro's (including the online version of office 356). I have tons of excel spreadsheets that do certain calculations for board design. Things i collected or wrote over decades. I even have excel psreadsheets that talk GPIB and collect data directly into graphs through the VBA engine.

Many companies are in that situation. Porting those is an impossible task. If only there was a way to run the VBA macros unmodified in google docs it would be the death of office. I don't know of anyone that uses VBA in word or the other office applications.
Office is used because of legacy Excel abased applications.

I ditched MS Access when i found out that you can use the free odbc driver with CSV files. No need for proprietary database formats. simple CSV tables. Anything in the same directory becomes a table for that driver. you can version them through GIT

[golden_labels]

Poland just released its own instant messanger for authorities: Narodowy Komunikator mSzyfr (≈ National Messanger mCipher). And authorities only, no access for the general public.⁽¹⁾

It’s advertised as “open source.” What is open source? Well… internally it relies on Matrix. Source is not available, the entire application is proprietary, apparently available only through a dedicated smartphone app, requiring Android and subscription to Google Store, and usage of Microsoft Authenticator.⁽²⁾ Authentication through other methods (e.g. Yubikey) is not supported. E2E is guaranteed through saying so in the info pages. Out of 17M PLN (≈$4.7M, 4M€) spent on developing it, no single cent went to the Matrix project/community. Matrix is not even attributed.

Did Poland do this because somebody believes FOSS is the way? No. Similar to Bundesheer it’s because they want to host it themselves. As opposed to the older solution, which was… oh, wait… on-premises, isolated version of Threema. Clients of which are all AGPL.

⁽¹⁾ In Polish: Portal Gov.pl, Spiderweb: Polski rząd chce pogonić WhatsAppa i Signala. Tym je zastąpi.
⁽²⁾ They don’t ban apps other than Microsoft Authenticator, but explicitly state they’re unsupported.

[magic]

The US citizenship is about 85-90% born in the US vs. naturalization. In Europe, the number is about 90% of the citizens born there. Not a huge difference, is it?

Another 10% of Americans don't even have citizenship and yet are claimed to be as good Americans as any other, at least by Americans. So that's what I find ironic, on one hand self-proclaimed group of free people who freely chose to live in free land and a "meaningless clump of cells" (Americans' own language referring to each other), on the other some grandiose "American nation". Get over it, you're a bunch of opportunists by definition and proud of it. As for EU being a wannabe US copycat, that's well known.

[magic]

It’s advertised as “open source.” What is open source? Well… internally it relies on Matrix. Source is not available, the entire application is proprietary, apparently available only through a dedicated smartphone app, requiring Android and subscription to Google Store, and usage of Microsoft Authenticator.⁽²⁾ Authentication through other methods (e.g. Yubikey) is not supported. E2E is guaranteed through saying so in the info pages. Out of 17M PLN (≈$4.7M, 4M€) spent on developing it, no single cent went to the Matrix project/community. Matrix is not even attributed.

Well, your (1) states that it's a replacement for Signal and WhatsApp.
Haven't they done a good job recreating those?

[golden_labels]

Well, your (1) states that it's a replacement for Signal and WhatsApp. (…)

It also doesn’t mention that Matrix is the underlying technology in mSzyfr. Unfortunately, for now you need to hope I do not lie.

[Jeroen3]

Earlier many of the programs didn't work right. But today everything is a browser window anyway.
Yes, even your "local applications" are often browser windows.

Added with the need to detach from the unreliable states of america, I can see this working.

Belgium also made their own messaging app iirc.

[golden_labels]

It’s not relevant that what Poland released is a messaging app. I believe you missed the point of that post, in the context of what is being discussed in this thread. And also what I said before.

[Gerhard_dk4xp]

.......you go by the Huron list, the US is second to China.

Greed for money and power is universal. The US certainly has problems, and hence the claim to be working towards "a more perfect union" is a goal. It's only been 250 years this year, you know. Some roads and buildings in other countries are older.

In our street here is a hospital that was founded in 1340.



Or, as a London woman said to an American:
"We should have been more careful rebuilding the town after the big fire!"

"?!!??"

"Yes, after 1666!"


Gerhard

[RoGeorge]

Oh yeah?!  Hold my beer!

Some are having iPad tablets now, but we used to have tablets about 7-8000 years ago:
https://en.wikipedia.org/wiki/T%C4%83rt%C4%83ria_tablets
 

[Nominal Animal]

I don't want to deal with figuring out how to cut a precise angle using the saw you provided.

That's just the end users.

It's the IT support that will be the real problem.  Think of them as your mentors or neighbors or colleagues that are supposed to be experts in the new equipment, and instead of telling you exactly how that cut could be done, they tell you you have to use a hand saw instead.  Or that it is just impossible, and you need to figure out something else.  Why? Because they just don't like the new equipment themselves; they too prefer the old comfortable stuff instead.

The power of open source comes from the support personnel.  In an optimal situation, they guide end users, but also monitor the needs and workflows of the end users, and communicate those to local developers, which make sure the tools support the most efficient workflows.

That is what I have done since '97 or so; as an IT support person, teacher, HPC sysadmin, and even for a couple of research groups as a researcher myself.  Without that, open source is just a zero-cost unreal mirage, a misconception.  With that, users making proverbial "cabinets" get the best tools possible for the tasks at hand.  I know, because I've done it myself, successfully, time and time again.  But the social friction and inertia is immense: it requires the humans involved to be well aware of how poor tools they now have.  (My strategy was to just cruelly show what could be possible first, to make them realize that.)

If you want to imagine what could be possible, think about the case where you get a plethora of small businesses that help anyone with not just Linux/FOSS IT support, but also help them optimize their workflows.  The companies have enough long-term vision to understand that staying small they stay nimble and competitive without huge administrative costs (most multinational companies are mostly administration), but co-operating (say with one company providing development and support personnel training, another providing support) they can compete with even the multinationals.  The problem with this is getting it going, and the fact that company growth will hurt their competitiveness; and that just does not gel with current investment and company-running strategies at all.  Today, the best you can do with Linux is purchasing a support agreement for the distribution you're using; forget any hopes of local support, unless you hire and train the support personnel yourself.

[Karel]

It's the IT support that will be the real problem.

The French gendarmerie managed to do it.

[Cyclotron]

Get over it, you're a bunch of opportunists by definition and proud of it. As for EU being a wannabe US copycat, that's well known.

I'm over it. Glad to hear we are living rent-free in your mind. 

[Nominal Animal]

It's the IT support that will be the real problem.

The French gendarmerie managed to do it.

And many companies around the world have managed to do it as well, yes.  Look at IBM, for example.  The Ministry of Justice in Finland has used OpenOffice for years now, too.

Do you understand what it means politically when an entire government tries to do it, with tens of thousands of workers currently enjoying pretty high-paid jobs required to completely retrain themselves or get fired?

In Finland, University of Helsinki failed to maintain even student-accessible Linux workstations because of this.  Around the turn of the century, they were common (as in entire classrooms full of Linux workstations) in the Faculty of Science (city center and Kumpula campuses), but they just couldn't keep them supported.  Students needed them, but as the IT support was separated to its own unit, they unilaterally chose to no longer support them, citing difficulty of hiring suitable support personnel IIRC.  The administration is too far above actual teaching and student needs to understand such decisions, even though they do the huge licensing deals (without actually knowing whether those deals make any sense).  The faculty instead decided to offer STEM students a laptop for a tiny bond sum, originally 100€, installed with locally-supported Linux, Cubbli.  I developed the first few versions of the self-registration Debian package, a GUI tool that generated local user accounts based on the University AD details over Eduroam, as a free volunteer, just to support the students and the Linux efforts.  The entire university has a bit over 31,000 students, 5,200 academic staff, and about 3,500 non-academic staff, so it's about third of the size of the French Gendarmerie.

In Finland, firing anyone from a government job or non-academic position in an University is almost impossible, due to the labour laws and extremely strong unions here.  "Quiet quitting" usually just leads them being shifted to another department or position, usually with same or higher pay but with fewer duties.  It's only taxpayer money, after all.  I wonder how many IT support personnel the Gendarmerie had to "shift", "retire", or fire.  I suspect it isn't easily done in France either.


One way this can work — noting that I'm not familiar with how the French Gendarmerie did it — is to motivate and train the support personnel first, years before the transition.  You need to go through the IT support administration, and carefully root out the change-fearing maintaintainers-of-status-quo, and basically force or brainwash the administration to be the proponents of the change.  (Maybe strongly worded; I do mean careful social engineering and political control.)  At the actual user-facing support personnel level, you must show how the change is in their own interests also.  And, at the end, those who won't perform in the new system, have to be fired or transferred.  I'd be very interested to hear how the Gendarmerie handled all that; it must have been a crucial part.

I know several computational physicists, who absolutely hate open source (calling it always "open sores"), even though they are capable of adapting software tools to their own needs, and occasionally have encountered bugs they had to fix themselves.  So, not just "fear" or "phobia" of any unknown thing; just very strong dislike of the entire idea.  When the same happened with commercial proprietary software, they are utterly nonchalant about it, because they understood that a commercial company must focus resources where they have the biggest bang for buck.  This kind of social, by definition irrational, attitude must not be ignored either.  Some are just fanboys and haters, regardless of their capabilities and skills.

[free_electron]

purchasing a support agreement for the distribution you're using.

You raise a valid point. But there are only few distros that come with paid support. Even less if you want to run certain applications. When Sun workstations and solaris started disappearing the switch to Linux started happening. The tool vendors like Mentor Cadence, Synopsis, Magma and others all pointed at Red Hat. They tested their products on very specific version of RHEL. If you installed it on any other distro you were totally on your own. If your design contains problems : sorry bub. we only guarantee and do bugfixes on this particular version.

More recently i ran into a problem with an Oracle install on CentOs. We had to migrate to something else. The problem was that either the Oracle engine wouldn't run on it, or, if we had a distro that would run the particular Oracle version, then the accessing application wouldn't run or be certified on that distro... Upgrading the oracle version was a no-go as the interfacing application needed a specific version and we were mid-design.
I've had the same problems with other tools like the Siemens teamcenter and NX. They need specific versions and specific distros. Try any "unapproved" combination and you are on your own. if it breaks or screws up : sorry bub, we don't test against that. you should have used xyz...

That is a big issue with linux: there are too many flavors. And commercial applications are typically tied to one particular distro and a few versions, especially if you want support for both linux and the running application.

[CatalinaWOW]

There are real problems in switching to a new ecosystem.  I have switched to LibreOffice for personal use, for two reasons.  First, its interface is more like the MS Office interface that I spent many years learning than the current MS Office interface.  Secondly, at the time I made the switch, MS Office was not available for purchase, only for rent, and I didn't want to do that.  For personal use that decision has worked fine, but I only have to satisfy myself.

LibreOffice does most tasks well, but as previously pointed out, it isn't compatible with MS automated documents.  I have a fairly sizeable investment in VBA functions to do engineering computations.  Things like probability density and inverse probability density for many distributions.  Blackbody calculations.  All out the window.  It is not obvious to me that I can replace this functionality without diving into the LibreOffice source code.  In any case substantial time would be involved learning the interface and then generating the functions.

The real problem is document interchange.   Complex documents created in MS don't come across well.  It is more than just "looks different".  Illustration titles can get separated from their illustrations.  Indexes scrambled.  It doesn't matter whose fault it is.  I don't know how broad this problem is, I can only report that I have encountered.  Are there problems between OpenOffice and LibreOffice?  As Nominal Animal points out even the distro may matter.

Issues like this are deadly to business and government.  Before I retired I often ran into requirements that proposals and bids be generated in a specific version of software, just to avoid these problems.

[Karel]

It's the IT support that will be the real problem.

The French gendarmerie managed to do it.

And many companies around the world have managed to do it as well, yes.  Look at IBM, for example.  The Ministry of Justice in Finland has used OpenOffice for years now, too.

Do you understand what it means politically when an entire government tries to do it, with tens of thousands of workers currently enjoying pretty high-paid jobs required to completely retrain themselves or get fired?

The French gendarmerie moved more than 100,000 stations to Linux. If they can do it at such a scale,
some ministry can do it as well. Somehow they managed to retrain their administrators (or fired them, I don't know).

[Nominal Animal]

That is a big issue with linux: there are too many flavors.

No, that's the perception stemming from assumptions and experience gathered from the proprietary software world.  I will fight you to death on this: you reduce the number of Linux flavors over my dead body.  I am a software developer myself, too, and not tied to any specific distribution.  I currently use two completely different ones: OpenWRT and a number of Debian derivatives.

It is perfectly possible to make even proprietary software that works on basically all Linux distros.  And I don't mean runtime images (Snap/Flatpak/Appimage), just a specific runtime directory and symlink configuration for associated libraries.  Why isn't it done?  Because it is different, and requires learning and quite some effort to understand how to do it and why.  Not worth it, because everyone expects software to either be distro-specific, or be provided as a runtime image.  It'd be pearls before swine to do that.  And yet, perfectly possible, although dbus and systemd-subsumed services do make it much harder than it used to be.

By reducing open source to what fits the traditional proprietary software view and approaches, you remove most of its true power.  It is like shooting yourself in the foot.  I do recognize that you want a working solution now rather than a better one sometime in the future, but these choices have long-term consequences.  To me, having seen and supported and used Linux in different environments, losing the variety is losing its utility.

This is also why I never recommend Windows users to switch to Linux.  They just don't get it, just like some of my computational physics colleagues don't get FOSS, even though they have all the requisite skills.  Too much to un-learn, to suddenly refuse to accept as a given fact, before one can even understand the new concepts.  And not enough reward at the end.

Just think: How many people will even accept that the variety in Linux implementations is its core strength, instead of pining for unification around a common distribution everyone would use?  Not many.  Portability was killed decades ago by proprietary software, and is now an utterly weird idea even to professionals.

[Nominal Animal]

The French gendarmerie moved more than 100,000 stations to Linux.

Taking two decades to do so.

If they can do it at such a scale, some ministry can do it as well.

Utterly ridiculous assertion, based on wishful thinking: you disregard both the political impact, and the time scales here.

Besides, the question is not whether it is possible, but whether it will succeed in practice.  As we know from U of H and Münich and others, there are a myriad ways how this can "fail", and not too many ways how it can succeed.

I do know how it could be made to succeed — by revamping the entire approach to computer use as a tool; including replacing courses on Word with courses on structured word processor use, and ensuring at every stage that it empowers the users and makes their workflows more efficient, never worse.  I also know, from practical experience, that it is the IT support personnel and administration that can make it succeed or fail.  And that means the issue whether such a transition will succeed or fail is not technical, but social and political.

[Cyclotron]

The French gendarmerie moved more than 100,000 stations to Linux.

Taking two decades to do so.

If they can do it at such a scale, some ministry can do it as well.

Utterly ridiculous assertion, based on wishful thinking: you disregard both the political impact, and the time scales here.

Besides, the question is not whether it is possible, but whether it will succeed in practice.  As we know from U of H and Münich and others, there are a myriad ways how this can "fail", and not too many ways how it can succeed.

I do know how it could be made to succeed — by revamping the entire approach to computer use as a tool; including replacing courses on Word with courses on structured word processor use, and ensuring at every stage that it empowers the users and makes their workflows more efficient, never worse.  I also know, from practical experience, that it is the IT support personnel and administration that can make it succeed or fail.  And that means the issue whether such a transition will succeed or fail is not technical, but social and political.

Taking that long and still not done. I read somewhere that it's about 95% complete, and that critical systems and issues with specialized and legacy software would require major projects funded in and of themselves to get transitions started. As I recall, the major issues were exactly what you've called out. Key challenges included managing the transition to a new interface, training staff, and ensuring compatibility.  These are the same challenges I recall reading that Germany faced when attempting the same sort of transition.

The key metric no one seems to publish is how much it costs them to work through these projects. They will often quote some number that they saved on license fees, but won't say what they spent to achieve the outcomes so far.  And there is the ongoing issue that they haven't actually achieved the goal since these critical systems haven't been migrated.

I still believe these are projects that governments can "afford" to pursue, but they sell them to their constituents and end up spending more without actually achieving the goal as the bureaucracy grows.

[free_electron]

requires learning and quite some effort to understand how to do it and why. 

and effort to test that it works correctly on all the variants... there is your problem.

A tool like Cadence Opus or Virtuoso (chip layout) is incredibly complex. They simply do not have the time nor manpower to try if it runs correctly without UI glitches ,font scaling issues and other stuff. Also the output generation needs to be correct. You can't have a problem where a tape out on RHEL produces different masks from running on ubuntu. Then there's also the deployment issue. this distro needs this installer, that one needs another..

Those tools are developed on a specific distro and they will only warrant it to run correctly on that distro. Deviate and you are on your own. Don't come crying if your tape-out doesn't work and you lost a multi-million dollar maskset and a fab cycle...

These are problems that don't exist on windows or MacOS. there is one kind of installer. There is also one UI server.

Anyway not my battle. i don't use use linux anymore. (stopped when we had that oracle problem when Centos was end of life. Moved the db to run on govcloud and ditched the local install and the oracle licenses. )

I make cabinets, not tools. I use screwdrivers, i don't make them. No time and / or desire to go into that rabbithole. An operating system should get out of my way. It;s there to read keyboard and mouse, manage the filesystem and run applications. If it can't run the applications i want to use, tough luck. I have a Mac and a PC with Windows. Serves me well. Decades ago i had a Sun Sparcstation as well but that ecosystem died and many of the tools were ported to Windows-only. Programs like Catia migrated from the unix world to the windows world.
Unix (not just linux) is simply too fragmented to test a program runs on all flavors. It is too much effort that is a distractor from the real work that needs to be done.

[tooki]

How much effort is worth investing in a task that cannot succeed? Especially when it goes against your own business interests?

IMHO a lot, since it's a Supplier & OS lock-in too.

I don’t think you understood my question?

You are correct here ... sorry i missed your point.

“A lot” means here “it is worth it for Microsoft to invest a lot in trying to make open file formats for Office, even though it can’t succeed and goes against its business interests”.

But doing so decreases lock-in.

I’m arguing that Microsoft didn’t have to make an active effort to increase lock-in, because merely investing the minimum in openness (as opposed to investing a lot) was more than sufficient to ensure lock-in.

Could be plausible.
MS Office has always been the "Golden Goose".  Not the one time "few bucks", made on a one time Windows OEM license.

Eh, the big bucks for MS historically was Windows licensing, because practically every non-Apple preassembled computer sold has a license. (Including the gazillions of computers used for specialty applications that weren't desktop office use.) But nowadays, the big bucks are from big corporate/government/education site licenses where they bundle everything together, including Windows, Office, and whatever else the customer needs.

[Nominal Animal]

and effort to test that it works correctly on all the variants...

Why?

If it is designed to be portable, there is no reason or need to test all the variants; simply that the facilities respond to random differences in the expected manner.  This is easily done in an automated manner.

The one stable interface is the Linux kernel - userspace one, including the /proc and /sys pseudofilesystems.  Userspace libraries (and select runtime services' like dbus and others in the systemd sphere) is where portability issues arise from.  So, for portability, you do not integrate those, and instead interface to a dynamic subset of them.

Those tools are developed on a specific distro and they will only warrant it to run correctly on that distro.

Yes, because that is how proprietary software is done, and how software development has been taught for the last two or three decades.  It does not mean it is the only way, or even the best way.  It is just what is the current way.  Some truly believe no other way exists, or could exist, because that is what they are used to.

I am not going to argue with you about this, because you are arguing about how things are done now, and I am arguing how things could be done better, but how social and political forces in specific non-user quarters make any change exceedingly difficult; and how one huge factor is exactly people used and fixed into the ways things are done now.  I fully agree with you how things work now!  And I'm content to work within those limits myself; I'm not the one proposing anyone else switching to Linux here or elsewhere!  I've never said it is easy or cheap to make software portable across Linux distributions, just that it is possible if desirable enough.  And, as many members here well know, I'm happy to help with proprietary problems and not divulge details to anyone, and am not a FOSS zealot at all.