Hacker News Reader

Was a post-mortem of the Munich project published anywhere?

Some people may remember the hype over the One Laptop Per Child Project, which had a noble mission that many people could get behind. In 2011 I heard from someone at MIT who criticized aspects of the program such as no real planning for support when the devices broke down in the field. Other issues came out years later, including unrealistic cost estimates and the crank:

If you remember the OLPC at all, you probably remember the hand crank. It was OLPC’s most striking technological innovation — and it was pure vaporware. Designers dropped the feature almost immediately after Negroponte’s announcement, because the winding process put stress on the laptop’s body and demanded energy that kids in very poor areas couldn’t spare. Every OLPC computer shipped with a standard power adapter.

https://www.theverge.com/2018/4/16/17233946/olpcs-100-laptop...

It's a shame it failed and there are no real cheap netbooks anymore (aside from Chromebooks).

I used an Acer Aspire One when travelling through Germany as a student (to research jobs and conferences) - and it was amazing being able to program and write papers on such a small device.

Reading about the hardware issues reminds me a lot of the ZX Spectrum reviews (especially the keyboard!) - keeping a low price means a lot of compromises unfortunately.

The modern equivalent is a tablet with its corresponding type cover.

AFAIK I can't really run Jupyter notebooks, R, LaTeX and Rust on a tablet though. Perhaps the JingPad.

You can on any Android tablet that allows root. You can run an X server via Termux and get a full graphical Linux environment. That will be necessary for Rust.

R, JupyterNotebooks and LaTeX can be done without root.

You can on any tablet that supports webassembly:

https://jupyterlite.readthedocs.io/en/latest/_static/lab/ind...

The GPD laptops fill that space for me now. My GPD Pocket literally fits in some of my pockets, is a regular x86 machine. Came with Windows 10, but I'm running Mint on it.

The problem is that a lot of organizations see foss as purely a free thing and not an investment. If they put half the money they spent on licensing into paying people to improve the foss packages they use, the ecosystem would transform and everyone would benefit. Massive cost savings overall, and it would push paid software to work harder to compete (who knows, it might even force Microsoft to offer decent privacy policies and remove ads from their manipulative desktop)

> If they put half the money they spent on licensing into paying people to improve the foss packages...

Like in a legally-binding contract way or some hippie way hoping for the best?

Hiring developers to work on specific package improvements is what is normally done. Could be legally binding, probably would be a standard employment agreement but you could do it through a hippie way (coding circle?).

great to see the "H-word" used as a slur, again (great for my own sense of antagonistic prejuidice being alive and well that is).

for a more constructive reinforcement of what a "hippy way" might be, in business and government:

anti-rascist, that is actively inviting and communicating with others socially; inclusive generally; valuing cooperation; building and valuing talent for its own sake; bringing in arts to shared spaces; generally vegetarian; health proactive; tolerance of personal drug use including tobacco; deeply environmentalist; generally opposed to military-style administration; generally opposed to authoritarian governments; use of non-monetary trade goods; valuing education and definitely advanced education.

are these things "fail" in City Administration as the parent comment implies ? not necessarily

"node_js_rulez_1" decries hippy open source support model.

... ummmmmmm ... look in your username.

Seriously, your stack is running on hippies, top to bottom.

TOP TO BOTTOM.

Organizations like school systems don't have a great track record managing custom software projects. They don't know what to look for and who to hire, and how much things should cost. It's very likely that these problems would translate into funding FOSS development as well. Also, it's a big headache to manage custom development compared to buying or downloading an off the shelf package.

This is the opposite of custom development for the most part, with the exception of some integration work.

It’s hard to predict whether you could make MS products in foss with global cooperation at lower cost to taxpayers (since it can be expensive to get politicians to agree with goals across international borders).

Actually, the Limux project is well alive: https://en.wikipedia.org/wiki/LiMux#Timeline

May 2020 - Newly elected politicians in Munich take a U-turn and implement a plan to go back to the original plan of migrating to LiMux.

Fascinating!

Wasn't that only initially, and was drastically improved later on? Even Accenture (the biggest Microsoft partner)'s audit didn't conclusively say that switching back to Windows will be better than what Limux was at that point.

Furthermore, the French Gendermerie is using Linux at a similar scale, and there are no usability problems. If they can make it work, i don't think it's impossible for anyone else.

> "Accenture (the biggest Microsoft partner)"

actually accenture's biggest partner is not microsoft, it's money. really, they would sell their employees souls for money.

I think the truth between what you say and the other person accusing you of being biased is somewhere in the middle. They basically went too mach ahead of the time. Mind you, the project started 18 years ago! The computing landscape was very different back then. I know what you mean as I was forced by my employer to use Linux on the desktop at that time and it was mediocre - but many things changed in the meantime. Probably the most important: the web revolution has happened. Nobody is using abominations like the ActiveX/Silverlight/whatever anymore. VBS is no longer omnipresent (partly because of security reasons). Hardware support in Linux is incomparable to what was in 2004. SO yes, I can imagine it was terrible back then, but I still think it's very important not to give up and I'm happy they basically restarted the project recently.

the hardware thing was not a problem for us as we could dictate which computers were bought. so computers were bought which would be compatible.

our problem was: none of the tools we needed were available in the quality and richness as they exist today.

automatic deployment to hundred of PCs at the same time with extensive automatic configuration? session and configuration management per user?

departments want to use different backgrounds per house or subdepartment. all that stuff had to be built in a scalable way.

the solutions today were not available back then.

I'm sorry, but I disagree about the two last points. Around 2001/2002 I was doing this kind of work of purchasing new desktops and deployments on hundreds of machines using a combination of PXE and Kickstart (I believe Kickstart appeared in 1998). It was not ideal, but it worked, and each department had their own settings. The only complaint I heard from my colleagues was that they cannot install "little sheep" (one of these gimmicks that would sit on your desktop and distract you). Actually some of them were upset they cannot install games, but they preferred not to voice this concern aloud.

As to your first problem, the lack of software, yes, it was a huge problem then. What we had to use at that time was substandard. I personally suffered (even though I love Linux on the server). Also KDE was unstable and a memory hog, Gnome wasn't much better, and you couldn't just buy more RAM. I'm very happy these days are over.

I always wondered why they had to create their own fork. I had the feeling that this is unecessary maintenance work that they took on and a "not invented here" philosophy.

Why not just use one of the avaible distros or why not even make a deal with SUSE (a bavarian company) to have a subscription model with support directly from experienced distro makers? Surely they could've made sure that all important features and workflows can be implemented.

Of course this would've needed to be a public bidding but if Red Hat gets it or SUSE does not really matter and surely is much cheaper than Microsoft licenses.

> I always wondered why they had to create their own fork. I had the feeling that this is unecessary maintenance work that they took on and a "not invented here" philosophy.

For the same reason there are hundreds of forks and counting: it is difficult to get the environment you want without rebuilding everything from scratch, and once you do that you have to maintain it.

They went with an Debian/Ubuntu based fork due to popularity.

>if Red Hat gets it

If your goal is to run a proprietary OS you might as well stay with Windows. No reason to endorse outdated IBM enterprise crapware after they killed the only compatible distro that didn't require paying the IBM tax to test and develop software for it.

The big distros are quite customizable in order to suit a large audience. A large organization typically has defaults and customizations that need to be added. A common strategy is to _distribute_ their customizations by creating a distribution based on some more widely used distribution. Google had Goobuntu based on Ubuntu.

because back in the day, no distribution was good enough for our needs. today is a completely other situation.

redhat might, but they borked it somehow as they said we would not be allowed to change anything or loose all support. at least that was what i heard.

and then there was supposedly the decision "we do not want any external company due to the bad experience with microsoft".

thanks for mentioning the Berlin airport fiasco - I had never heard the story - hilarious and very troubling for those in Berlin!

Oh, if you find the story of BER hilarious and troubling you should read up on the 2021 vote for federal and local parliaments. Or the civil administration there in general where you have to make an appointment to get your passport renewed months in advance.

> 2021 vote for federal and local parliaments

tldr?

Some people are unhappy with the results.

One candidate was generally hated by the public and got fucked by the press with "creative photo editing", badly.

One candidate was poised to become the first green chancellor, but then managed to have not one, not two, but three case of academic fraud and lying about her past tacked to her ticket. Her party came in second.

One candidate was all about some nebulous idea of "freedom", which made him relevant enough that his party now is king maker, when the first-time voters, who traditionally would go to the Green Party, decided enough Covid Lockdowns is enough.

And one candidate did ... nothing. Absolutely nothing. Which means he also couldn't embarrass himself (even though historically, he's a ... problematic ... figure). This guy came in first, so he's going to be the next chancellor.

And that's just the candidates. Then you got the State of Berlin obviously unable to hold democratic elections, given

- there was a lack of ballots in many voting offices

- in some voting offices, there were ballots for other districts, which rendered those votes invalid

- they accidentally allowed 16-year-olds to vote in the federal elections (where you have to be 18 to vote)

- they accidentally gave a mandate to the wrong guy, who shared the same name, but not the same party affiliation with the actual winner

- some votes were cast after 18:00, which is the official cutoff date

- in some districts, there was a 150% participation rate

- some offices sent people waiting in line home because "it's unlikely you'll be in in time"

- ... I am sure I am missing a few catastrophes - a LOT went wrong in Berlin ...

It was hilarious to watch this mess unfold.

>...I REALLY hope the LibreOffice transition goes well.

If anything that rings alert on my mind is the LO's Calc, which at times decides to become slow and consume the CPU for no apparent reason. Or equally annoying flacky interoprability betwen LO Writer and MS Office Word (works in general, but loses some alignment, magles styles etc.)

As much as I'm for LO and want them to succeed, I feel for the clerks dealing with these counter-productive issues.

You say you want them to succeed, but I don't believe you are sincere. This negativity with anecdotal evidence is so annoying. Like the whole world will come to a halt because some document formatting got fucked up.

I see this same shit with Excel all the time. It's fucking FULL of bugs that have just been around for so long, people just get used to it and work around it.

I mean, if you selected a group of 30 rows in Excel, then use CTRL to deselect one row in the middle... it doesn't deselect it, it "selects" the row again, so now it's just darker than all the other rows. Is this fixed yet? It was a bug in fucking Office 2016 and goes back years. Nobody said it wasn't ready for use!

It's such a fucking basic feature. To de-select a single row out of a group that is selected. It's the single most counter-productive thing ever. When I found this bug, I was blown away that nobody ever complained about it! I worked with thousands upon thousands of people with Excel and it was never a problem?!

Word doesn't even play nice with it's own files... ever moved a file from Office 2013 to Office 2016? I've seen it garble files a million times. Move a Word doc from Mac to Windows and shit gets hosed sometimes.

The clerks will get used to the issues. Just the same way that they got used to the hot fucking garbage that Microsoft charges enterprises and governments substantial amounts of money for, and provides precisely dick for support in return.

EDIT: I mean look at this shit! Apparently deselecting is a new concept to Microsoft!

https://support.microsoft.com/en-us/office/deselect-a-select...

Note: This feature is only available in Excel for Window if you have Office 2019, or if you have a Microsoft 365 subscription. If you are a Microsoft 365 subscriber, make sure you have the latest version of Office.

>... You say you want them to succeed, but I don't believe you are sincere.

Not sure what drives your experience, but I've been using LO (Linux based) continuously for the past 10 yrs. Still do. More so, tried to convert other users. I don't believe I'm the only one to stumble on some usability issues with LO.

Did I get used to dealing with the issues? Nope. This still annoys me every time, it's also a drag to interoperate with MS Office (no, it's not going away soon in this world). Other users? Well, most of my attempted converts now either pay the subscription or use the GSuite or ... stick to the older version of MS Office.

So I myself do want the LO to get better, but for the most of the failed "converts" it's a lost cause for now.

> This negativity with anecdotal evidence is so annoying.

Isn’t it.

> “It's such a fucking basic feature. To de-select a single row out of a group that is selected. It's the single most counter-productive thing ever.”

Probably not as counter-productive as throwing away thousands of person-years of learned skill and experience because of some Linux fanaticism.

How is “it’s not as good but we can patch it until it’s equivalently buggy, people will get used to it” any compelling reason to change over?

How is “but but I hate microsoft” any compelling reason to change over?

And the techcommunity.microsoft thread from 2018 announcing deselect has people who used that “multi-click makes a cell darker” feature and are bothered that it’s gone.

> “Microsoft charges enterprises and governments substantial amounts of money for, and provides precisely dick for support in return.”

You mean like that feature you were crying out for and then found that they built? That kind of nothing?

Or like how Office 365 has collaborative editing and LibreOffice has a page saying it’s been in development since 2006 and isn’t ready and linking to an old mailing list post from 2020 talking about deprecating he API they had built for it, as the latest update?

I suspect the reason for that is nobody ever tried a Linux deployment as big as that, so the tooling to manage it simply didn't exist.