I think the definition of FOSS used here is tendentious: some of these projects (which I have no particular attachment to) are marked as "not FOSS" or "issues exist" because they have components that are disconnected from the basic nature of free and open source software itself.
A recurring one here seems to be that proprietary builds somehow make a project not FOSS. But this is how it's always worked: Red Hat doesn't sell FOSS source, they sell a subscription to a distribution (RHEL) that includes managed, maintained builds. That distribution is in turn restricted[1], while the source behind it remains free.
Perhaps there's an argument to be made that the definition of FOSS should be stronger, and should include some kind of binary freedom, lack of trademark restrictions, etc. But that's not how the term is conventionally applied, and glossing over that convention seems roughly as contentious as when companies try to split the baby and rewrite "open source" to include anti-competitive terms.
In those situations, could someone easily just fork the project, offer builds, and now their version of the project is ideal? If it's easy to do that then it seems like a good ideal. If it is difficult to do then their right it is an 'issue'.
One to add: NanoKVM is definitely fake FOSS. It dials home to download a serialized, opaque library. NanoKVM: The S stands for Securityhttps://youtu.be/plJGZQ35Q6I
Many people also seem to think Atlassian Jira and Confluence are OSS when they're absolutely not.
They also used to offer source downloads to paying customers, but never claimed to be open source, because the licence they gave to customers to access that source did not allow for distribution. (It was meant for auditing, and help in building extensions, I think)
F-Droid, the FOSS-centric alternative app store for Android, provides similar information for each app. F-Droid goes a bit farther on things mobile users care about, like calling out if an app sends telemetry or requires a paid subscription. I like that this Is it really FOSS? project examines a project's potential impact on the FOSS community by questioning whether it is VC-funded, requires a CLA, and other interesting characteristics like that.
F-Droid goes way off the deep end on other issues, though. For instance, most RSS readers get the "non-free network services" warning because you can use them to follow RSS feeds on any web site, including ones whose software isn't open-source.
Do you have an example? Because of the 10 RSS reader I just checked only a couple have that warning. It's also not about being able to add non free sites (if so all browser would have that warning) but that they come by default with them.
Some entries are at best confusing, and at worst misinforming.
The common case is considering projects which have one element that is FOSS and another that isn't. For example: ProtonMail, who apparently offer a FOSS mail client. They never presumed to offer mail server software; and FOSS mail server software is available. So a button calling them out for not being really FOSS kind of misses the mark. You don't see an entry like that for, say, GMail - so if Proton did not provide a client at all, they would have faired better.
Another specific case is that of Signal. The client and server are FOSS, but they're designed for no federation, so you can't (?) use a modified Signal client with the vanilla clients, and you definitely can't add a server to the network. This effectively prevents modified versions of Signal from being usable. So, is it really FOSS? The site's verdict is: Unqualified yes, Green button.
> ProtonMail, who apparently offer a FOSS mail client. They never presumed to offer mail server software
The website justifiably cites this website marketing copy as misleading: "All Proton services are open source and independently audited for security." https://proton.me/mail If that's supposed to only apply to the mail client (which isn't specifically mentioned on that page), it's incredibly unclear.
Indeed. Glad it actually cares more Free aspect than the OS aspect. It's important to speak truth to power - VCs are abusing open source as distribution channel and then doing a rugpull, and those who care about FOSS need to fight back in these ways.
The real test of "freedom" isn't just the license but whether users can realistically deploy modified versions in production without prohibitive technical or legal barriers.
I got banned permanently from this server (with no explanation of why). However, it look like it is also on Codeberg, which I am not banned from (although many of the links are incorrect when viewed on Codeberg (at least if JavaScripts are disabled), it still works).
I think these articles are good, but I do have some other comments.
For some programs, there is the possibility that some parts can potentially work without non-FOSS but is difficult to separate. (This can also be a different problem in case you only want one part of the program.)
A program can also be Free but "trapped", in case it requires proprietary compilers to compile it (although it is often possible to work around this; sometimes easily and sometimes more difficult).
For some games that have non-FOSS parts, there is also the issue of if the non-FOSS parts can execute arbitrary code or otherwise do things outside of the game itself, that is not necessarily desirable (e.g. a Game Boy Advance emulator might be FOSS, although the programs it emulates might or might not be FOSS, but either way do not affect the rest of the computer nor the internet and other stuff like that); and, also the consideration of whether the software can be used without the non-FOSS parts (if you can replace them; e.g. a FOSS game engine might be made as a clone of a non-FOSS game engine that can use the original game files but you can also make your own fully FOSS games using it too).
There is also some that may require non-FOSS to access, even if the software itself is FOSS. Proprietary (or overly complicated, even if FOSS) communication channels are also not mentioned (although another comment on here does mention it), and I think it probably is a concern (not one that necessarily makes the project itself to be not FOSS, but still might be worth mentioning), even if it does not make the program itself to be not FOSS, it can make it difficult to contribute or to use it.
Being FOSS does not necessarily mean that you intend to run the program on your computer; you might only want to view the code, or modify it before running it, or use your own program (or a different FOSS program) as a substitute.
Programs can be "open core" but the non-FOSS part is still clearly distinct from it (which is the case for SQLite). (In the case of SQLite, they also mention the non-FOSS test suite; they are not needed to run the program, but it may make it difficult to make your own changes and then test it. However, some programs do not have a real test suite at all, anyways.)
This project's source code is hosted on Codeberg, which runs on the FOSS forge Forgejo.
This is a big improvement over projects that are hosted on GitHub. For those, the license may be FOSS, but the spirit is not, because anyone that wants to contribute upstream is lured onto a proprietary platform.
The license and terms of service of a project's community communication channels are not listed under the concerns. (https://isitreallyfoss.com/concerns/) This is understandable: traditionally and strictly, the license is the only thing that matters.
A whole lot of FOSS folks don't like that the vast majority of folks do not care about license distinctions that ultimately won't ever be tested in court anyway.
Ultimately, most people use the term "open source" to mean "freely downloadable" or similar. Sorry I guess that the gnulag[1] never happened.
Most people won't care if food and medicine are FSA/FDA certified or not, and most people won't care if a business is SOC certified. But the few people that do care about it make the world's medicine better, and the businesses better.
I think this is not particularly impartial. Sentry is marked as "NOPE" even though it is basically open source (any commit older than two years is), yet projects that are open core forever are "issues exist" and "partially".
> Sentry is marked as "NOPE" even though it is basically open source (any commit older than two years is),
In other words, current Sentry isn't Open Source, but old versions are available. I think it's a fair characterization that Sentry is not Open Source, unless there's an actual community around the Open Source version.
I agree with the first part ("current Sentry isn't Open Source"), not with the second one ("unless there's an actual community"). A software can be Open Source without actual development community.
Ah, wait -- you don't say that Open Source requires a community. You state that, if there was interest in the older version of Sentry, then that version could be considered as Open Source. I agree with this one as well, then.
[I considered deleting this comment, but maybe it will be useful to others who have not read closely.]
Right, exactly. I'm not saying a project must have a community in order to be Open Source. I'm saying in order to be considered for a list like this, something should be a project in its own right rather than just a stale version of another project.
Having a community is not necessary for licenses to have a certain effect. I made the argument in the past that the FSL gives you more rights than the AGPL, it just does it slightly later. The effect however is greater. See https://lucumr.pocoo.org/2023/11/19/cathedral-and-bazaaar-li...
The FSL, like the BuSL that Hashicorp Terraform is now under, is a proprietary software license.
"Partially" makes sense for something where some component or subproject is released under a proprietary license and other components or subprojects are released under a free software license.
Software that will eventually be open-source isn't yet open-source. A once-F/OSS codebase that was closed and removed from GitHub two years ago also has all commits that are 2+ years old open-source, but that that doesn't make such a codebase open-source, either.
I don’t agree with this assessment though. By that logic if I just happen to release everything I do with a two year delay it’s open source by the logic of this website but it would have the same effect.
You don't have to agree, but it doesn't have the same effect.
The reality is the company is releasing software to customers that have no way of using the source code to that software under an open source license. You can claim this is somehow open source because they will eventually be able to use the source code under an open source license, but that doesn't make the software open source today, any more than you can lock someone in jail and claim they're actually free because they'll get out in two years.
Why is this project switching its license 2 years ago during development not a problem? What is good about Sentry? edit: Oh you mean rolling time windows.
Or, at least, not giving a fair take on its particular approach to open source.
We both self-host and pay for the service. There is ample engagement from the development team with the larger community. There are also a myriad of open source projects without the same licence restrictions that Sentry-the-company publishes or maintains which make up key functionalities in Sentry-the-product.
Given that it's almost been 2 years since the FSL change[1], I would imagine it's possible to set up a rolling mirror into https://gitlab.com/glitchtip/sentry-open-source/sentry that serves as an easier "what is actually open right now?" browser.
I just realized while typing that out: a snapshot of the source code actually doesn't help anyone know if a particular commit is Apache 2 since the change is dependent upon an accounting record on GitHub
> The project is licensed under an FSL-1.1-Apache-2.0 license which, for two years after release, prevents use, modification and distribution when done in a range of ways which may compete with the original project.
The current license for sentry seems to be a large part of the reason for the nope.
They give a pretty detailed explanation of the decision.
It's easy to blame evil companies for attempting to monetize OSS, it's harder to accept that a lot of the reason for more company focused OSS is that indie OSS devs were historically treated poorly, not just by companies but also by entitled users within the OSS community. A poignant example years ago was "devs" with empty GitHub commit histories coming into the OSS community to harass small projects into adopting their badly made and legally untested codes of conduct (and then attacking the individuals running those projects when they pushed back).
When you're not being paid to do something, the only benefit you get aside from software you use yourself is friendly peer recognition, and when it becomes too abrasive, when people are treating you like politicians and trying to scare you into adopting their political views, when users come in and trash talk your project like they're your boss because you didn't implement some feature they want, a lot of people just give up and leave. I largely left the space because of this, and a lot of really good OSS contributors I knew did too.
I'm not sure what the solution is at this point but it's probably not a continuation of the entitlement mentality, purity tests and witch hunts that this site is perpetuating.
One one hand it's probably a good thing to have open source police. On the other hand, not quite open source was the correct choice for a lot of these projects.
Not open source is a perfectly fine choice, as long as people are honest about it (and they are complying with licences if its based on other open source projects). The problem is when they lie about it or try and pull a bait and switch.
Pretty sure it's impossible to not use closed source Google or Apple pieces for a functional app these days. Last I looked in to it, the only way to actually deliver notifications is to run through Google or Apple since the OSs don't want every app running their own background processes draining battery.
I think this is really unfair, in the current day and age, especially when there are "Open Weight Models" that are bending the definition of the FOSS.
I don't have a skin in the game, but I personally think that the definition of FOSS is too rigid and strict and is not evolving. There has been many challenges over time (LGPL's linking exception, tivoization, AGPL trying to fight against SaaS, Open Core business models, ...); and we are really bestowing very harsh moral standards for people who are trying to do the right thing.
For me, Sentry, being 10+ years in its existence (I used it ever since its logo was a Starcraft II unit), never participated in the usual enshitification of the software, being labeled as "NOPE" is disingenuous. I would gladly pay for Sentry because I love the software, and I also know that if shit hits the fan, I can self-host it (though the configuration for self-hosting got progressively difficult over time, but that's the complexity of modern SaaS stack). I can make similar arguments to other tools in this site that I'm familiar with.
FOSS is indeed a moral and ethical stand on freedom, than on business realites (that's more why "open source" came about).
You can't have the FOSS cake and eat it too.
For eg: Sentry can release sentry-open under a fully free license from moral considerations, but choose not to do so because of business considerations. That's an OK choice to make, but you hence don't get to call yourself.
Its ok for things not to be open source, but for the term to mean anything it has to have a definition. Vauge moral handwaving about "doing the right thing" doesn't really help give meaning to the term. A project can be open source and morally bad. It can be closed source and morally good. Authors' moral intentions are a totally orthogonal dimension.
Even if it solved its own energy crisis? Could it ever unquestionably deserve sovereignty? At the very least on git pull requests as an individual global contributor?
If LLM output is unlicensed, maybe an LLM could one day be trained on the public output of every existing LLM. A second gen LLM that can confidently identify its own kind alongside public domain human ideas.
I think you’re being downvoted because not everybody on HN knows that Llama is not open source, despite Yan LeCunn ignoring the OSI and continually attempting to tell everyone it is.
My wife is Venezuelan, and when they think something smells they say Foss. It's a never ending source of amusement when I'm browsing hn. She will love this site.
I think the definition of FOSS used here is tendentious: some of these projects (which I have no particular attachment to) are marked as "not FOSS" or "issues exist" because they have components that are disconnected from the basic nature of free and open source software itself.
A recurring one here seems to be that proprietary builds somehow make a project not FOSS. But this is how it's always worked: Red Hat doesn't sell FOSS source, they sell a subscription to a distribution (RHEL) that includes managed, maintained builds. That distribution is in turn restricted[1], while the source behind it remains free.
Perhaps there's an argument to be made that the definition of FOSS should be stronger, and should include some kind of binary freedom, lack of trademark restrictions, etc. But that's not how the term is conventionally applied, and glossing over that convention seems roughly as contentious as when companies try to split the baby and rewrite "open source" to include anti-competitive terms.
[1]: https://www.redhat.com/en/resources/red-hat-enterprise-linux...
In those situations, could someone easily just fork the project, offer builds, and now their version of the project is ideal? If it's easy to do that then it seems like a good ideal. If it is difficult to do then their right it is an 'issue'.
… isn’t that basically what CentOS did in the early days?
What happened? Why did it stop being easy and some people say "CentOS Linux is dead—and Red Hat says Stream is “not a replacement”?
If the company no longer makes it easy that's a big issue. On a list like this, CentOS would appear next to Red Hat Linux Enterprise.
Oracle Linux?
One to add: NanoKVM is definitely fake FOSS. It dials home to download a serialized, opaque library. NanoKVM: The S stands for Security https://youtu.be/plJGZQ35Q6I
Many people also seem to think Atlassian Jira and Confluence are OSS when they're absolutely not.
Atlassian stuff never was, they do offer free licenses to use their cloud products to open source projects though - perhaps that's the confusion?
They also used to offer source downloads to paying customers, but never claimed to be open source, because the licence they gave to customers to access that source did not allow for distribution. (It was meant for auditing, and help in building extensions, I think)
The vid is old and they made some promise to open source things later.
I wonder if it ever happened. I did withhold my purchase back then just because I'd rather wait for open source than buy some device I cannot trust.
"Old"? 5 months? Are you some sort of impatient consumer? ;oD
Who is "they"? NanoKVM or the reviewer?
There appears to be a follow-up comparison video here: https://youtu.be/CsfB_Avi2-4
F-Droid, the FOSS-centric alternative app store for Android, provides similar information for each app. F-Droid goes a bit farther on things mobile users care about, like calling out if an app sends telemetry or requires a paid subscription. I like that this Is it really FOSS? project examines a project's potential impact on the FOSS community by questioning whether it is VC-funded, requires a CLA, and other interesting characteristics like that.
F-Droid goes way off the deep end on other issues, though. For instance, most RSS readers get the "non-free network services" warning because you can use them to follow RSS feeds on any web site, including ones whose software isn't open-source.
That's not accurate. Feeder is the only RSS app that has a warning like that, and it's because it talks back to feeder-sync.nononsenseapps.com.
Do you have an example? Because of the 10 RSS reader I just checked only a couple have that warning. It's also not about being able to add non free sites (if so all browser would have that warning) but that they come by default with them.
https://f-droid.org/en/docs/Anti-Features/#NonFreeNet
"apps that promote or depend entirely on a proprietary network service."
Some entries are at best confusing, and at worst misinforming.
The common case is considering projects which have one element that is FOSS and another that isn't. For example: ProtonMail, who apparently offer a FOSS mail client. They never presumed to offer mail server software; and FOSS mail server software is available. So a button calling them out for not being really FOSS kind of misses the mark. You don't see an entry like that for, say, GMail - so if Proton did not provide a client at all, they would have faired better.
Another specific case is that of Signal. The client and server are FOSS, but they're designed for no federation, so you can't (?) use a modified Signal client with the vanilla clients, and you definitely can't add a server to the network. This effectively prevents modified versions of Signal from being usable. So, is it really FOSS? The site's verdict is: Unqualified yes, Green button.
> ProtonMail, who apparently offer a FOSS mail client. They never presumed to offer mail server software
The website justifiably cites this website marketing copy as misleading: "All Proton services are open source and independently audited for security." https://proton.me/mail If that's supposed to only apply to the mail client (which isn't specifically mentioned on that page), it's incredibly unclear.
Ok, fair enough, but then - you would expect isitfoss to counterpose the claim with the reality.
They have "Details" for each project doing exactly that: https://isitreallyfoss.com/projects/proton-mail/
You missed how there are five possible answers for a binary yes/no question. The site is confusing by design.
https://isitreallyfoss.com/about/categorisation/
You can add a new project to the website by creating an issue [1].
[1] https://codeberg.org/danb/isitreallyfoss/issues
oh good a purity test for OSS projects, this is exactly what's needed in the ecosystem, and will surely have a positive impact
Indeed. Glad it actually cares more Free aspect than the OS aspect. It's important to speak truth to power - VCs are abusing open source as distribution channel and then doing a rugpull, and those who care about FOSS need to fight back in these ways.
i am getting the sense that you have missed the point
(Good, i guess make your point simply and directly, so it isn't turned to the opposite case)
The real test of "freedom" isn't just the license but whether users can realistically deploy modified versions in production without prohibitive technical or legal barriers.
I got banned permanently from this server (with no explanation of why). However, it look like it is also on Codeberg, which I am not banned from (although many of the links are incorrect when viewed on Codeberg (at least if JavaScripts are disabled), it still works).
I think these articles are good, but I do have some other comments.
For some programs, there is the possibility that some parts can potentially work without non-FOSS but is difficult to separate. (This can also be a different problem in case you only want one part of the program.)
A program can also be Free but "trapped", in case it requires proprietary compilers to compile it (although it is often possible to work around this; sometimes easily and sometimes more difficult).
For some games that have non-FOSS parts, there is also the issue of if the non-FOSS parts can execute arbitrary code or otherwise do things outside of the game itself, that is not necessarily desirable (e.g. a Game Boy Advance emulator might be FOSS, although the programs it emulates might or might not be FOSS, but either way do not affect the rest of the computer nor the internet and other stuff like that); and, also the consideration of whether the software can be used without the non-FOSS parts (if you can replace them; e.g. a FOSS game engine might be made as a clone of a non-FOSS game engine that can use the original game files but you can also make your own fully FOSS games using it too).
There is also some that may require non-FOSS to access, even if the software itself is FOSS. Proprietary (or overly complicated, even if FOSS) communication channels are also not mentioned (although another comment on here does mention it), and I think it probably is a concern (not one that necessarily makes the project itself to be not FOSS, but still might be worth mentioning), even if it does not make the program itself to be not FOSS, it can make it difficult to contribute or to use it.
Being FOSS does not necessarily mean that you intend to run the program on your computer; you might only want to view the code, or modify it before running it, or use your own program (or a different FOSS program) as a substitute.
Programs can be "open core" but the non-FOSS part is still clearly distinct from it (which is the case for SQLite). (In the case of SQLite, they also mention the non-FOSS test suite; they are not needed to run the program, but it may make it difficult to make your own changes and then test it. However, some programs do not have a real test suite at all, anyways.)
This project's source code is hosted on Codeberg, which runs on the FOSS forge Forgejo.
This is a big improvement over projects that are hosted on GitHub. For those, the license may be FOSS, but the spirit is not, because anyone that wants to contribute upstream is lured onto a proprietary platform.
The license and terms of service of a project's community communication channels are not listed under the concerns. (https://isitreallyfoss.com/concerns/) This is understandable: traditionally and strictly, the license is the only thing that matters.
A whole lot of FOSS folks don't like that the vast majority of folks do not care about license distinctions that ultimately won't ever be tested in court anyway.
Ultimately, most people use the term "open source" to mean "freely downloadable" or similar. Sorry I guess that the gnulag[1] never happened.
[1] https://i.redd.it/th4prtdk6xr61.jpg
That's true everywhere.
Most people won't care if food and medicine are FSA/FDA certified or not, and most people won't care if a business is SOC certified. But the few people that do care about it make the world's medicine better, and the businesses better.
I think this is not particularly impartial. Sentry is marked as "NOPE" even though it is basically open source (any commit older than two years is), yet projects that are open core forever are "issues exist" and "partially".
> Sentry is marked as "NOPE" even though it is basically open source (any commit older than two years is),
In other words, current Sentry isn't Open Source, but old versions are available. I think it's a fair characterization that Sentry is not Open Source, unless there's an actual community around the Open Source version.
I agree with the first part ("current Sentry isn't Open Source"), not with the second one ("unless there's an actual community"). A software can be Open Source without actual development community.
Ah, wait -- you don't say that Open Source requires a community. You state that, if there was interest in the older version of Sentry, then that version could be considered as Open Source. I agree with this one as well, then.
[I considered deleting this comment, but maybe it will be useful to others who have not read closely.]
Right, exactly. I'm not saying a project must have a community in order to be Open Source. I'm saying in order to be considered for a list like this, something should be a project in its own right rather than just a stale version of another project.
Having a community is not necessary for licenses to have a certain effect. I made the argument in the past that the FSL gives you more rights than the AGPL, it just does it slightly later. The effect however is greater. See https://lucumr.pocoo.org/2023/11/19/cathedral-and-bazaaar-li...
It's in no way "basically open source", and the list strikes me as completely impartial. You cite Sentry, which says this on their licensing page:
> All components powering the main Sentry and Codecov web apps use FSL, which limits their usage in a commercial Sentry-like offering.
The license limits usage in a commercial context. That's not FOSS by the OSI's definition, which is the definition I care about.
Heck, you can tell from the URL that Sentry _really_ wants to pretend they are open source.
https://open.sentry.io/licensing/
The FSL, like the BuSL that Hashicorp Terraform is now under, is a proprietary software license.
"Partially" makes sense for something where some component or subproject is released under a proprietary license and other components or subprojects are released under a free software license.
Software that will eventually be open-source isn't yet open-source. A once-F/OSS codebase that was closed and removed from GitHub two years ago also has all commits that are 2+ years old open-source, but that that doesn't make such a codebase open-source, either.
I don’t agree with this assessment though. By that logic if I just happen to release everything I do with a two year delay it’s open source by the logic of this website but it would have the same effect.
You don't have to agree, but it doesn't have the same effect.
The reality is the company is releasing software to customers that have no way of using the source code to that software under an open source license. You can claim this is somehow open source because they will eventually be able to use the source code under an open source license, but that doesn't make the software open source today, any more than you can lock someone in jail and claim they're actually free because they'll get out in two years.
Wouldn't "partially" be fair? Since not ALL of the project is, but only source of a certain age?
That in my mind would be a much fairer categorization.
Why is this project switching its license 2 years ago during development not a problem? What is good about Sentry? edit: Oh you mean rolling time windows.
Or, at least, not giving a fair take on its particular approach to open source.
We both self-host and pay for the service. There is ample engagement from the development team with the larger community. There are also a myriad of open source projects without the same licence restrictions that Sentry-the-company publishes or maintains which make up key functionalities in Sentry-the-product.
How often does Sentry contribute to its own two year old branch? Is anyone maintaining a fork of it that would deserve to be on this list?
I don't believe it's a fork, but https://gitlab.com/glitchtip/glitchtip-backend is MIT and aims to be the actual Open Source replacement for Sentry
Given that it's almost been 2 years since the FSL change[1], I would imagine it's possible to set up a rolling mirror into https://gitlab.com/glitchtip/sentry-open-source/sentry that serves as an easier "what is actually open right now?" browser.
I just realized while typing that out: a snapshot of the source code actually doesn't help anyone know if a particular commit is Apache 2 since the change is dependent upon an accounting record on GitHub
That was the advantage of the BuSL 1.1 scheme, that those cut-over dates were embedded in the file https://github.com/getsentry/sentry/blob/22.7.0/LICENSE#L19 and https://github.com/getsentry/sentry/blob/22.8.0/LICENSE#L19 is open in 10 days
1: https://github.com/getsentry/sentry/commit/d4d89e2fa93ad99f1... is Nov 17, 2023 -05:00
> The project is licensed under an FSL-1.1-Apache-2.0 license which, for two years after release, prevents use, modification and distribution when done in a range of ways which may compete with the original project.
The current license for sentry seems to be a large part of the reason for the nope.
They give a pretty detailed explanation of the decision.
It’s still a good product.
It's easy to blame evil companies for attempting to monetize OSS, it's harder to accept that a lot of the reason for more company focused OSS is that indie OSS devs were historically treated poorly, not just by companies but also by entitled users within the OSS community. A poignant example years ago was "devs" with empty GitHub commit histories coming into the OSS community to harass small projects into adopting their badly made and legally untested codes of conduct (and then attacking the individuals running those projects when they pushed back).
When you're not being paid to do something, the only benefit you get aside from software you use yourself is friendly peer recognition, and when it becomes too abrasive, when people are treating you like politicians and trying to scare you into adopting their political views, when users come in and trash talk your project like they're your boss because you didn't implement some feature they want, a lot of people just give up and leave. I largely left the space because of this, and a lot of really good OSS contributors I knew did too.
I'm not sure what the solution is at this point but it's probably not a continuation of the entitlement mentality, purity tests and witch hunts that this site is perpetuating.
One one hand it's probably a good thing to have open source police. On the other hand, not quite open source was the correct choice for a lot of these projects.
Not open source is a perfectly fine choice, as long as people are honest about it (and they are complying with licences if its based on other open source projects). The problem is when they lie about it or try and pull a bait and switch.
Signal I suppose is open source, but doesn't it contain closed source Google binary blobs? It also routes through closed source notification libraries
Pretty sure it's impossible to not use closed source Google or Apple pieces for a functional app these days. Last I looked in to it, the only way to actually deliver notifications is to run through Google or Apple since the OSs don't want every app running their own background processes draining battery.
Practical concerns must not matter for a purity test.
On iOS that's basically true, but on android there are a lot of battery preserving options
I think this is really unfair, in the current day and age, especially when there are "Open Weight Models" that are bending the definition of the FOSS.
I don't have a skin in the game, but I personally think that the definition of FOSS is too rigid and strict and is not evolving. There has been many challenges over time (LGPL's linking exception, tivoization, AGPL trying to fight against SaaS, Open Core business models, ...); and we are really bestowing very harsh moral standards for people who are trying to do the right thing.
For me, Sentry, being 10+ years in its existence (I used it ever since its logo was a Starcraft II unit), never participated in the usual enshitification of the software, being labeled as "NOPE" is disingenuous. I would gladly pay for Sentry because I love the software, and I also know that if shit hits the fan, I can self-host it (though the configuration for self-hosting got progressively difficult over time, but that's the complexity of modern SaaS stack). I can make similar arguments to other tools in this site that I'm familiar with.
FOSS is indeed a moral and ethical stand on freedom, than on business realites (that's more why "open source" came about).
You can't have the FOSS cake and eat it too.
For eg: Sentry can release sentry-open under a fully free license from moral considerations, but choose not to do so because of business considerations. That's an OK choice to make, but you hence don't get to call yourself.
Its ok for things not to be open source, but for the term to mean anything it has to have a definition. Vauge moral handwaving about "doing the right thing" doesn't really help give meaning to the term. A project can be open source and morally bad. It can be closed source and morally good. Authors' moral intentions are a totally orthogonal dimension.
> For me, Sentry, [...], being labeled as "NOPE" is disingenuous
How is it disingenuous? The current version of Sentry is proprietary.
Free but free to sustain? No way. Sometimes there are companies that accompany a paid, Enterprise version (HashiCorp, GitLab).
You get what you pay for
What about the singularity? When AI can maintain any program?
Even should that occur, AI isn't and wouldn't in that case be free of cost.
Even if it solved its own energy crisis? Could it ever unquestionably deserve sovereignty? At the very least on git pull requests as an individual global contributor?
Now add an "open source" LLM.
If LLM output is unlicensed, maybe an LLM could one day be trained on the public output of every existing LLM. A second gen LLM that can confidently identify its own kind alongside public domain human ideas.
I think you’re being downvoted because not everybody on HN knows that Llama is not open source, despite Yan LeCunn ignoring the OSI and continually attempting to tell everyone it is.
https://opensource.org/blog/metas-llama-2-license-is-not-ope...
https://opensource.org/blog/metas-llama-license-is-still-not...
downloadable ≠ open source.
https://www.downloadableisnotopensource.org/
Yes. That's what I'm saying (more specifically, I am saying the OSD defines open source).
[dead]
My wife is Venezuelan, and when they think something smells they say Foss. It's a never ending source of amusement when I'm browsing hn. She will love this site.