Page 1 of 1

[Core Developers] Minetest's migration from GitHub

PostPosted: Wed Sep 04, 2019 09:06
by Hamlet
This thread is meant to let the Minetest's community know if the Minetest Core Developers are willing to move from GitHub to a Free and Open Source git hosting platform.

[EDIT]
2019-09-06
The current position of the Minetest's Core Developers team can be read in these posts:
viewtopic.php?f=3&t=23243#p356293
viewtopic.php?f=3&t=23243#p356322
viewtopic.php?f=3&t=23243#p356391
[/EDIT]

Minetest is a Free and Open Source project, thus some community members believe that it would be coherent moving either to an already existing platform which adheres to the Open Source filosophy, or to a Minetest self-hosted platform.

Please reply to this thread either by stating "yes" or "no", and if you wish add an explanation for your answer.

This thread has been notified by email to:

celeron55
sfan5
ShadowNinja
Nore/Ekdohibs
nerzhul/nrz
sofar
rubenwardy
Krock/SmallJoker
Lars

By private message to:
paramat

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Wed Sep 04, 2019 18:33
by sfan5
We've had this discussion on Github: https://github.com/minetest/minetest/issues/7412
You will be able to find (or infer) stances of most devs in the discussion there.

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Wed Sep 04, 2019 18:38
by Krock
You might also be interested in/or have posted to following topics:
Emigrate from Github? (Jun 2018)
Are we not worried that Minetest is depending on MS now? (Sep 2019)
What Git (or other) repo are people migrating to? (Sep 2018)

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Wed Sep 04, 2019 20:01
by Hamlet
@Sfan5, @Krock

Actually "I" (those who are interested to this matter) would prefer a direct answer to a direct question - if this does not mean to ask too much - instead than being sent to this or that thread (some of which I have been part of) to fathom the Core Developers' opinions, "I" might misunderstand their position.

A "yes", a "no", a "yes but not now" and a "no, never" can not be misunderstood.

Furthermore, Sfan, that issue report is one year old, therefore some opinions are obsolete, example:
paramat: "I expect MT will leave, but it will probably be at least 6 months before that happens or has to happen." (2018-06-27)

Only 5 core developers out of 10 have expressed their position.

Please, be nice and reply "yes" or "no" to this thread, so that we can have an up to date feedback, and people not using Git can always look at this thread for the purpose that it is meant for: knowing your final word.

If a "yes" or a "no" seem too simplistic, you can always add a line like "read here for my full opinion [link]".

Thank you.

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Wed Sep 04, 2019 20:19
by sofar
The question you pose is ambiguous. You say:

if the Minetest Core Developers are willing to move


But since this is unbound, you are going to get scattercloud answers based on different presumptions (move, if when?). It also is problematic that you use the words "willing to" and not something like "agree to move".

I would be (read full paragraph until the end first!) OK with a move permitted that there is sufficient reason to warrant a move. I think everyone agrees with that exact statement, because it is clearly bound, we can just debate the meaning of "sufficient" here (but I'll leave that debate to Wuzzy ;)).

So there is no way in hell that you are going to get a "yes/no" answer from anyone. That would be incredibly unfair anyway. In these sort of discussions, it is critically important that everyone is allowed to express their full opinion. Only allowing "yes" or "no" is completely the opposite of fairness. It's also not how a community like MT works.

There's nothing wrong with regularly having this discussion. If you approach it like an engineering problem, then it's not hard to see that "not moving" is just as valid as "moving" of a solution to the perceived problems. Everything comes at a cost. Moving has benefits but major downsides. Not moving has other downsides. If we regularly (not every week, maybe not every year either) revisit the choice and carefully consider all the aspects, we're doing exactly what should be done.

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Wed Sep 04, 2019 20:34
by ShadMOrdre
Perhaps more appropriate questions? Hard coded yes/nos not required, but the more specific an answer, the less likely it can be misinterpreted, misunderstood, or otherwise not clearly spelled out.

Would the developers state any specific requirements they might have for any potential platform, so that viable options can more easily be determined?

IF the community, or someone within it, can and will provide a viable alternative to using github, OR if a viable alternative can be determined, will the developers consider using that option? (Take consider to mean not just consider, but an actual commitment to make the move...)

IF it is decided to migrate, will the developers announce a time frame or move by date?

On a different note...

Is there not an Open Source repository somewhere? Where is the official linux kernel hosted? Linus made git.

It's hard to fathom that the OS community dropped the ball and hasn't risen to the challenge here, leaving it instead for proprietary companies to fill the void. One would think there'd be an OS git, that ALL OS projects should be using, instead of all the closed-source source code repositories.

Shad

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Wed Sep 04, 2019 23:06
by Hamlet
@Sofar, @ShadMOrdre

Please let's try - for once - not to hijack the thread, nor to turn it into an endless and useless - for the community - debate. We have plenty of them, don't we?
This thread is a poll for the Core Developers, not to argue about its correctness.
I am a simple person, I have made a simple question, I am fine with a simple answer.

The question is very clear and simple:
1 - Minetest is OpenSource software
2 - GitHub is not an OpenSource platform
3 - Are you [core developer] willing to move Minetest's development to an OpenSource git platform? / Do you [core developer] want to move Minetest's development to an OpenSource git platform?

Yes - no further explanations needed
No - optional explanations previously expressed [link]
Yes but not now - optional explanations previously expressed [link]

Please refrain from turning this thread into an argument between you and me: it does not help the community, nobody cares about another endless debate/argument.
If you think that my question is wrong for whatever reason, just ignore it.

Let me stress it: this thread is a poll for Core Developers, to let the community know what they think about migrating Minetest's development from GitHub.

The question stands.

Thank you.

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Wed Sep 04, 2019 23:35
by sofar
If there is a reasonable conclusion that migrating is needed or significantly beneficial to Minetest and outweighing the costs, then I'm 100% sure that the decision to move would be backed by all the developers.

Logic dictates that because this move hasn't happened or isn't urgently being discussed, that this is not the case right now, and so, to answer presumptuously for ALL the core developers, everyone votes "no, but we'll keep it in mind", which I think you should have included as a reasonable option. It is fundamentally different as an option than "yes but now now", since that suggests that a move is going to happen. Why?

It's kinda like Brexit - once you announce you're leaving you can't really stop. I'm turning it around and saying that it would be a whole lot smarter to say "we're staying for now, but we'll keep entertaining the cost/benefit equation" than to go all Tory and "we're leaving, but, we have no clue as to what damage it does when we do and we have no plan to even make it happen".

You asked how I think, here it is.

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Thu Sep 05, 2019 09:01
by Hamlet
sofar wrote:If there is a reasonable conclusion that migrating is needed or significantly beneficial to Minetest and outweighing the costs, then I'm 100% sure that the decision to move would be backed by all the developers.

Logic dictates that because this move hasn't happened or isn't urgently being discussed, that this is not the case right now, and so, to answer presumptuously for ALL the core developers, everyone votes "no, but we'll keep it in mind", which I think you should have included as a reasonable option.
[...]
You asked how I think, here it is.


Thank you for your answer.

Thus - please correct me if I am wrong - the fact that GitHub is not OpenSource has no weight for your "vote", nor for the other Core Developers?

I am asking because that is the reason of this poll, and the Minetest project is already on https://gitlab.com/minetest

Am I correct in assuming that for "you" (Core Developers) moving on GitLab would be negative (not significantly beneficial) and/or it would cost more than what the project could gain (not outweighing the costs)?

Post Scriptum:
Whatever your answer to my last questions, I will be satisfied.
If no other core devs reply saying something different, I will consider your position the official position of the core devs team.

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Thu Sep 05, 2019 09:38
by Linuxdirk
sofar wrote:If there is a reasonable conclusion that migrating is needed or significantly beneficial to Minetest and outweighing the costs, then I'm 100% sure that the decision to move would be backed by all the developers.

We're in the time of memes.

Image

This is very shortened and very shallow, but also painfully accurate.


sofar wrote:Logic dictates that because this move hasn't happened or isn't urgently being discussed, that this is not the case right now

c55 said that the change will happen if Microsoft "breaks something". So I guess we just need to wait.

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Thu Sep 05, 2019 12:36
by Xudo
Hamlet wrote:Thus - please correct me if I am wrong - the fact that GitHub is not OpenSource has no weight for your "vote", nor for the other Core Developers?

IIRC GitHub never was open-source. Even before acquisition by Microsoft they were commercial company with closed source of their backend.
So yes. I think the fact that GitHub is not open source has no weight for anyone who is hosting their open-source project there.

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Fri Sep 06, 2019 06:18
by sofar
Hamlet wrote:Thus - please correct me if I am wrong - the fact that GitHub is not OpenSource has no weight for your "vote", nor for the other Core Developers?


We discussed this when it happened. While some people pulled all their repos off of github, aside VanessaE nobody else appears to have done so. That seems to reasonably support that statement, so yes, I do not thing that most of the core devs think that it is worth moving over. Something significantly worse has to happen.

Hamlet wrote:I am asking because that is the reason of this poll, and the Minetest project is already on https://gitlab.com/minetest


Having a shadow project at a different hosting place is smart. It does not mean that the central repository is moved. It's a backup. We've already had it. Nerzhul ran a copy for years prior to this on gitlab.

Hamlet wrote:Am I correct in assuming that for "you" (Core Developers) moving on GitLab would be negative (not significantly beneficial) and/or it would cost more than what the project could gain (not outweighing the costs)?


All moves are bad. Moving away from github is really, really bad. Moving to gitlab is also really, really bad. Not a single move will be a net positive. We already lost developers that no longer could fathom github. There will be developers that won't be able to swallow gitlab, either. Thus, the least harmful thing right now that can be done is to ... not move.

"3.6 roentgen, not great, not terrible" -> you can't assess the cost of something you don't know about.

There is something to be said for self-hosted. This is realistically the only real solution in case github goes bonkers, since, if github goes bonkers, it's quite likely that gitlab also does (they're a company run by people, after all, and who says that github won't "assimilate" them either, for instance) soon thereafter. So, any move would have to significantly weigh this risk and assure that we're not moving to another unstable vessel. Hence, self-hosted is really the only safe place to go to. It ironically would give the most flexibility to cater to developers, but also would have a significant human cost in the form of maintenance. Again nothing fun to look forward to.

Hamlet wrote:Post Scriptum:
Whatever your answer to my last questions, I will be satisfied.
If no other core devs reply saying something different, I will consider your position the official position of the core devs team.


Oh, you can totally consider that! I mean, I wouldn't do that, but, I can see your point - there's very little to add to this discussion at this point.

TL:DR; yeah, a move is unlikely. no, we're not ignoring it. yeah, nobody likes the alternatives. no, moving won't be difficult, but let's not right now?

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Fri Sep 06, 2019 06:24
by sofar
Linuxdirk wrote:Image
This is very shortened and very shallow, but also painfully accurate.


It's entirely OK to be wary, make sure you don't get weary!

Tech companies are like multi-headed dragons. One side will actively be killing something cool, while the other side just approved a native EXFAT driver for the Linux kernel that isn't even made by them!

I still won't put Windows on my systems, though. Don't take me for a fanboy either.

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Fri Sep 06, 2019 07:23
by Linuxdirk
sofar wrote:... while the other side just approved a native EXFAT driver for the Linux kernel that isn't even made by them!

Oh no, don't bring up this second state of their EEE policy. I follow Microsoft's business practice since I got aware of it and were old enough to start caring about such things (must have been around the early days of Windows 98 if I remember correctly). No matter if Gates, Ballmer or Nadella, they all aim for the same goal with different approaches. Nadella clearly read Sun Tzu's The Art of War and tries to keep "the enemy" close.

Sorry, but I am not rational when it comes to Microsoft because of what they did the last 30 or so years.

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Fri Sep 06, 2019 09:54
by ShadMOrdre
We can all complain about the big corps. M$ isn't the only beast out there. Whether one is for/against, fanyboy or fanatically opposed, I simply want stuff that works, when I want it to work. Unfortunately, for the better part of the last 50 years, M$, Intel, the Wintel monopoly that still is a monopoly, and now the crapple, spamdroid, inyerfacebook, and snoople are tearing down the remaining walls of privacy, decency, and launching the beginnings of skynet.

We go out, not with a wimper, but with a market cap of bazillions because people can't get enough of this abuse.

We'd all be better off directing our angst to action, and towards the right directions. Tech company abuses pale in comparison to big oil, big pharma, big food, and big medicine.


Shad

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Fri Sep 06, 2019 12:15
by Hamlet
@sofar

Thank you for you clear and detailed answer, finally we all know what we wanted to know. I have consequently edited the first post.

This thread will be useful for future reference for anyone who wishes to know the Core Developer's position, or to open yet another thread about git hosting alternative solutions - such a thread will be useful for its participants, but "the last word" will always be available here.

I believe that everything has been asked and that each question had its answer, thus I will not add anything else.

Thanks for your availability and helpfulness, it is much appreciated. :)

@forum moderators
If you wish, feel free to lock this topic: as "poll" has achieved its goal.
Perhaps "pin it" in the "sticky" section, so that everyone can easily find it.
- just a suggestion, of course -

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Fri Sep 06, 2019 23:06
by paramat
Am i willing?
In the future: Yes, but only if necessary in a purely practical way. Currently: No, because there have not been any very bad purely practical changes yet.

By 'purely practical' i mean:
I disagree with migration being influenced by idealogical, political, philosophical aspects, including 'open source software' versus 'proprietary'.
I don't much like Microsoft, but i disagree with migration being influenced by the fact that Microsoft is the owner, let's wait until they actually do bad things with GitHub (if they ever do).

So my attitude is similar to celeron55's previously stated opinion.

Do i want to?
No, i hope it is never necessary to do so, as migration is disruptive and we have very limited time, and GitHub is good.

I wrote "I expect MT will leave, but it will probably be at least 6 months before that happens or has to happen."
You can see i was half-expecting things to go bad quite quickly, i have been pleasantly surprised. It's possible that MT never has to leave, hopefully.

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Thu Nov 21, 2019 06:54
by FreeGamers
Hamlet, thanks for making this thread. Its a topic I'm interested in and I wrote some thoughts about this in another thread here.

Today, I created my first remote git repo I've ever made and chose to do it on notabug.io. I felt and still feel a lot of pressure to use GitHub because most people use it and a lot of projects use it. However, I don't want to be a victim of Microsoft's marketing choices in the future. Nor do I want to encourage others to join that community to work on my code with me. I think the philosophy of "it works fine now, so why change it" is short-sighted. I appreciate this thread because it is very simple, direct, and informative about where the core-devs stand on this type of issue and what type of philosophies are present in the project's core.

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Sat Nov 23, 2019 11:36
by Astrobe
sofar wrote:All moves are bad. Moving away from github is really, really bad. Moving to gitlab is also really, really bad. Not a single move will be a net positive. We already lost developers that no longer could fathom github. There will be developers that won't be able to swallow gitlab, either. Thus, the least harmful thing right now that can be done is to ... not move.

"3.6 roentgen, not great, not terrible" -> you can't assess the cost of something you don't know about.

There is something to be said for self-hosted. This is realistically the only real solution in case github goes bonkers, since, if github goes bonkers, it's quite likely that gitlab also does (they're a company run by people, after all, and who says that github won't "assimilate" them either, for instance) soon thereafter. So, any move would have to significantly weigh this risk and assure that we're not moving to another unstable vessel. Hence, self-hosted is really the only safe place to go to. It ironically would give the most flexibility to cater to developers, but also would have a significant human cost in the form of maintenance. Again nothing fun to look forward to.


It appears to me that the answer has been given - "yes, but not now" is the safest answer - I think I can go slightly off topic.

It seems to me that people sometimes confuse Git with the Git-based Web services. I have heard the expression the other day of "social features" about some of the GitHub new stuff. Personally I took it as a bad omen because when something goes "social", it means "attractive for the masses" and this is where thing usually start to go awry. Nobody in their right mind would go after the masses just for the love of the crowds; you usually want a critical mass of users so that e.g. your RTS players can find a good opponent without waiting in queue too much. When you go after the masses, you are really after the money it carries, like a pickpocket in the crowd. But I digress.

Git itself is a command-line tool at the core. There's a GUI utility on the side, there's a server utility (git-daemon), but there's no Web GUI. Yet you can exchange commits via the Git protocol (understood by git-daemon) or by email (git-format-patch) or any text messaging device that does not "nicely" reformat your message (which is unfortunately the case of many email client software). More conveniently, you can push your changes from the command line and almost never see the ugly face of GitHub or Gitlab or whatever.

So if/when you move away from GitHub, and if self-hosting is out of the equation, then one should consider a minimalist host. Because my guess is that you don't lose developers because of Git, but because of the -Hub or the -Lab part. That is, the extra features such as issue tracking, wikis, avatars and whatnot. Managing the interactions between contributors is a separate concern which is best done with specialized software.

True, it is less convenient. Exchanging commits right here on this forum is actually an option. Pull requests, reviews, etc. as well. It would of course probably the most inconvenient of all options. More likely you'd want at least software specialized in issue tracking.

But it is also more resilient to accidents such as your service provider going terrible, evil or under. The success of Git, that follows the single responsibility principle which is part of the Unix philosophy (do one thing and do it well), has shown the value of it.

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Sun Nov 24, 2019 19:46
by paramat
FreeGamers wrote:Nor do I want to encourage others to join that community to work on my code with me. I think the philosophy of "it works fine now, so why change it" is short-sighted.

Our approach is a rational and practical approach with no fear, prejudice, assumptions or hysteria.
Although MS owns Github, Github is not MS, it is a mistake to treat it as such, this is not really 'direct association' with MS.
I can completely understand not wanting to host your work on Github, but by leaving Github completely, a contributor actually harms Minetest far more than harming MS, because it becomes more difficult to comment on development or submit contributions.

In the world of business there are many buyouts, the owning company can buy a company to own it as an asset, but the bought company often keeps the same staff and often continues much as before. The ownership may be rather 'hands off', as i get the impression MS liked Github the way it was.

The best approach to Github is: "I am not too happy about your new owner, but you are still Github and i will stay with you as long as your owner does not make a mess of you".

To be clear: I rather dislike MS, and as far as i know the other core devs feel similarly.

Re: [Core Developers] Minetest's migration from GitHub

PostPosted: Mon Nov 25, 2019 00:42
by Linuxdirk
paramat wrote:I can completely understand not wanting to host your work on Github, but by leaving Github completely, a contributor actually harms Minetest far more than harming MS, because it becomes more difficult to comment on development or submit contributions.

Actually I see only one solution here, and that solution is self-hosting an instance of whatever “collaborative Git web-frontend” devs think is the best one. So when Microsoft fucks up GitHub and the switch is indispensable, maybe use that once-in-a-lifetime opportunity to consolidate all development activity fully under own control.

Yes, this will change things, but changes happen, and changes are not always bad. And if GitHub will finally be left what’s the difference in changing to git.minetest.net or commercial GitLab, or BitBucket? Self-hosting a web application for Git is only hard when setting it up. Maintenance isn’t really that different from maintaining other web-based applications.