There are too many PRs

User avatar
Hume2
Member
Posts: 631
Joined: Tue Jun 19, 2018 08:24
GitHub: Hume2
In-game: Hume2
Location: Czech Republic

Re: There are too many PRs

by Hume2 » Post

"use an editor which allows you to jump between functions"
You can also use an editor which shows you more that 80 characters per line :D

@Zughy: If you were one of the core dev, do you think that you could reinforce the team?
If you lack the reality, go on a trip or find a job.

User avatar
Linuxdirk
Member
Posts: 2741
Joined: Wed Sep 17, 2014 11:21
In-game: Linuxdirk
Location: Germany
Contact:

Re: There are too many PRs

by Linuxdirk » Post

Hume2 wrote:
Mon Jun 15, 2020 16:51
You can also use an editor which shows you more that 80 characters per line :D
But who in the right mind wants that?

User avatar
Hume2
Member
Posts: 631
Joined: Tue Jun 19, 2018 08:24
GitHub: Hume2
In-game: Hume2
Location: Czech Republic

Re: There are too many PRs

by Hume2 » Post

Linuxdirk wrote:
Mon Jun 15, 2020 17:10
Hume2 wrote:
Mon Jun 15, 2020 16:51
You can also use an editor which shows you more that 80 characters per line :D
But who in the right mind wants that?
Do you see the paradox? Having an editor which can search the definition is obvious but an editor which displays more than 80 chars per line is a huge problem.
If you lack the reality, go on a trip or find a job.

User avatar
Zughy
Member
Posts: 276
Joined: Thu Mar 26, 2020 18:23
GitHub: belongs_to_microsoft
In-game: Zughy
Location: Italy
Contact:

Re: There are too many PRs

by Zughy » Post

Hume2 wrote:
Mon Jun 15, 2020 16:51
@Zughy: If you were one of the core dev, do you think that you could reinforce the team?
I'm nowhere near having the core devs' knowledge, I couldn't reinforce anything even if I wanted right now. The best thing I could do is being a moderator on the forum as I open it every day and I report as much as I can (also I know 4 languages, so I can help also where some other admins can't), but:
1) I've been pondering it for a while as I don't want to overwhelm myself (see previous point #3)
2) I trashed that possibility with my French in the previous message, which I'm not sorry for because I felt my (and my friend's) time not taken seriously because of the same old story

I think I'm more useful developing mods (which I really enjoy) and managing "stuff"

CalebJ
Member
Posts: 339
Joined: Sat Dec 29, 2018 19:21
GitHub: KaylebJay
IRC: KaylebJay
In-game: CalebJ
Location: Tunnelers' Abyss

Re: There are too many PRs

by CalebJ » Post

Zughy wrote:
Mon Jun 15, 2020 13:25
... as there is no official documentation, not even a file describing the hierarchy of it or a few guidelines.

There is the lua_api.txt for modders, but I presume you are talking about documentation for the engine. Yes, I agree it could use documentation. Its a matter of someone making it and submitting a PR, which if you are true about your code style accusations, will take quite a bit of revisions and time to make.
Zughy wrote:
Mon Jun 15, 2020 13:25
Just try to imagine how much time they're willing to sacrifice in order to understand the code: they'll be gone before saying "hello". And this is YOUR fault, not theirs.
What's with all the ad-hominem attacks of the core devs? Face it: the core devs spend a lot of their time doing something that is completely for others. And now you're telling them to go redo the whole minetest engine because some people can't understand the code, and then blaming them for these people leaving? <insert Hume2's signature here>
Zughy wrote:
Mon Jun 15, 2020 13:25
2) You're terribly disorganised. Like, for real. Wuzzy spent enough words about it, so let's keep the perspective from an external point of view only: 160+ PR, 950+ issues, milestones saying nothing about priorities. What are we going to see in 5.4.0? Or in 5.5.0? Will there actually be one or it'll go directly to 6.0.0? Where is the roadmap? The short answer is: there is none. Sure, there is a document saying MT goal is to be as generic and optimised as possible, but that's it. What happens right now is "huh, this seems important. Yeah, let's add it to the next release... for now". The end.
Organization is in the eyes of the beholder. Look at it this way: We have all our PRs and issues in one place and many collaborators and a few devs to chat about them and get them fixed/merged/closed. That's organization. What isn't organization is this: https://github.com/minetest/minetest/issues/9989 ;) Secondly, we do have milestones, as you mentioned. But the features of future versions - it doesn't work that way in Open Source unless you have some super-coder. It works that way in games where you have hundreds of paid developers and they get everyone excited about new features the next versions will bring. <insert Hume2's signature here>
Zughy wrote:
Mon Jun 15, 2020 13:25
3) Some of you lack focus, massively wasting time and energies. I can't stress this enough, starting from MTG. You're complaining you don't have enough dev time, yet you keep developing and supporting a modding base advertised as a (horrible) game. I'm sorry if someone might feel offended, but that's what MTG looks like (have a look at the survey, that's not me saying it).
MTG is meant to be a base-game that mods are put on. Yes, I agree, its not a game. But that doesn't mean we can't play it. (I did in singleplayer for quite a while, looking for diamonds and making underground railways and farms mostly - before trying out servers) Maybe the devs could be wasting their time with MTG, but it is the base game and the silent majority/minority would complain a lot if they stopped developing it or took it down. (besides, the latter would break almost every mod :D)
Zughy wrote:
Mon Jun 15, 2020 13:25
But let's remain on topic: there is, what, 4 of you? 5? Start cleaning that huge mess that are MT issues, because it looks evident you can't support both the engine and the game, so why keep bothering? Also, rubenwardy is actively following more projects than three normal persons put together.
Nothing against rubenwardy, but I think the other core devs have useful opinions, make useful PRs, work with people to try to understand/fix issues, and try to get things merged (including minetest-mods and MTG), as well as Ruben.

The core devs are only human, Zughy. If you want development done faster, attacking core devs won't help a thing. If they didn't care about minetest, they wouldn't be here, so be grateful they are spending their free time working to further FOSS and Minetest.
Zughy wrote:
Mon Jun 15, 2020 13:25
I'm nowhere near having the core devs' knowledge, I couldn't reinforce anything even if I wanted right now. The best thing I could do is being a moderator on the forum as I open it every day and I report as much as I can (also I know 4 languages, so I can help also where some other admins can't)
So ... you're attacking someone when you've never been in their shoes. And you don't even want to be in their shoes - this is likely because their work is something that a large community is watching and analysing. This is akin to slamming a baseball manager for doing something dumb, when you have absolutely no experience with baseball or managing a baseball team. Hindsight is always 20/20 ;)
If you want realism, you're in the wrong place. Get off your mobile.

User avatar
Zughy
Member
Posts: 276
Joined: Thu Mar 26, 2020 18:23
GitHub: belongs_to_microsoft
In-game: Zughy
Location: Italy
Contact:

Re: There are too many PRs

by Zughy » Post

And now you're telling them to go redo the whole minetest engine
Wait what? I asked to write a documentation, I didn't mention redoing the engine at all
the core devs spend a lot of their time doing something that is completely for others
Last time I checked I did too. Maybe you skipped the part about being a volunteer: read again. Also, that's not an excuse to do something badly
Secondly, we do have milestones, as you mentioned.
Please show me
it doesn't work that way in Open Source unless you have some super-coder
This sounds like an excuse
If you want development done faster, attacking core devs won't help a thing
Last time I checked the one whose bugs wouldn't have been noticed if Clobber hadn't insist for his PR (creating a few problems in the 5.3.0) was me. Also, it's pretty hypocrite to say when you're shitting me with the Hume2's signature
If they didn't care about minetest, they wouldn't be here
Yeah, me neither. I waited 3 days before posting this thing. If I didn't care, I wouldn't have done it, trust me
so be grateful they are spending their free time working to further FOSS and Minetest
I've never stated I'm not: their work is admirable. I stated they're doing it badly, both for them and for people who want to help. So if they ask me how to improve it, I'm entitled to share my opinions
you're attacking someone when you've never been in their shoes. And you don't even want to be in their shoes
No, this is about being realistic. If both my C++ and my MT engine knowledge is poor, why would I even want to be in their shoes? For what, for creating some mess because I don't understand basically anything? Because I want to be a leader? Yeah, no thanks
when you have absolutely no experience
I'm criticising their managing skills explaining them why they won't find any new help and why this thing will get worse, I'm not criticising their coding skills. And I definitely have experience with the former (EDIT: I wrote "latter", my bad). Also, this is illogic: it's like saying art critics aren't entitled to criticise any artist till they don't reach the artist level. Like, uhm, no?
Last edited by Zughy on Tue Jun 16, 2020 13:57, edited 1 time in total.

CalebJ
Member
Posts: 339
Joined: Sat Dec 29, 2018 19:21
GitHub: KaylebJay
IRC: KaylebJay
In-game: CalebJ
Location: Tunnelers' Abyss

Re: There are too many PRs

by CalebJ » Post

Zughy wrote:
Mon Jun 15, 2020 19:52
And now you're telling them to go redo the whole minetest engine
Wait what? I asked to write a documentation, I didn't mention redoing the engine at all
You said the code was confusing. If it is true as you say, it will mean huge code revisions will have to be made before a documentation can be made that makes sense.
Zughy wrote:
Mon Jun 15, 2020 19:52
the core devs spend a lot of their time doing something that is completely for others
Last time I checked I did too. Maybe you skipped the part about being a volunteer: read again. Also, that's not an excuse to do something badly
Most of us are volunteers, Zughy, including myself (not for minetest though). It doesn't mean we are automagically granted the audacity to dumb-down other volunteers. Sorry :)
Zughy wrote:
Mon Jun 15, 2020 19:52
Secondly, we do have milestones, as you mentioned.
Please show me
https://github.com/minetest/minetest/milestones
Zughy wrote:
Mon Jun 15, 2020 19:52
it doesn't work that way in Open Source unless you have some super-coder
This sounds like an excuse
Not quite, because I'd only be making an 'excuse' if I was a core dev defending myself.
You can't have it both ways: you can't hate core devs but want as much progress and development as possible.
Last time I checked the one whose bugs wouldn't have been noticed if Clobber hadn't insist for his PR (creating a few problems in the 5.3.0) was me. Also, it's pretty hypocrite to say when you're shitting me with the Hume2's signature
ClobberXD is not a core dev, btw. O_o
Yeah, me neither. I waited 3 days before posting this thing. If I didn't care, I wouldn't have done it, trust me
Good to know.
I've never stated I'm not: their work is admirable. I stated they're doing it badly, both for them and for people who want to help. So if they ask me how to improve it, I'm entitled to share my opinions


*If* this was true, then I wouldn't say their work was admirable. Yes, you are entitled to share your opinions, but I only saw one reasonable criticism in your whole post - about lacking documentation. Wuzzy wrote up a beautiful criticism with valid points and his best thoughts for improving the situation. You seem to just be tongue-lashing core devs for absolutely no *good* reason. Rather you should suggest improvements and give constructive criticism.
No, this is about being realistic. If both my C++ and my MT engine knowledge is poor, why would I even want to be in their shoes? For what, for creating some mess because I don't understand basically anything? Because I want to be a leader? Yeah, no thanks
Hmm, so you're admitting you would make a mess of it and you also admit to your MT engine knowledge being poor. However, you also think your opinions will make minetest better. o_O
when you have absolutely no experience
I'm criticising their managing skills explaining them why they won't find any new help and why this thing will get worse, I'm not criticising their coding skills. And I definitely have experience with the latter. Also, this is illogic: it's like saying art critiques aren't entitled to criticise any artist till they don't reach the artist level. Like, uhm, no?
Again, you can criticize all you want. The validity of your 'arguments' is what is being questioned.

Trash-talking core devs means, for one, you are not grateful for their long work over the years, but two, you are ruining this topic. Wuzzy did not blame core devs at all, but tried to reach out and give them ideas on how to improve (which I largely agree with). The core devs are donating a large part of their time to us players either way, so although I do hope they are taking Wuzzy's suggestions seriously, I also hope they do not take your trash-talking seriously. If you want change, make a PR or reach out to the devs. You are not doing that in any way here.
If you want realism, you're in the wrong place. Get off your mobile.

User avatar
sfan5
Moderator
Posts: 3962
Joined: Wed Aug 24, 2011 09:44
GitHub: sfan5
IRC: sfan5
Location: Germany

Re: There are too many PRs

by sfan5 » Post

Zughy wrote:
Mon Jun 15, 2020 13:25
When I expressed my disappointment for the PR system, sfan5 answered me with the same old refrain of "it's disappointing [...] but how do you want to fix the issue of dev time?"
Yes, and it's true. It's not like your rant isn't justified or not accurate, but the question remains.
I personally put an amount of time I consider appropriate into MT, depending on motivation and also because there is no point in burning yourself out.
But even if I wanted to, I cannot run engine development on my own. There always needs to be a second developer that also reviews the same PRs.

Zughy wrote:
Mon Jun 15, 2020 13:25
You're complaining you don't have enough dev time, yet you keep developing and supporting a modding base advertised as a (horrible) game.
If you look at https://github.com/minetest/minetest_ga ... its/master you will see that no significant amount of developer time has gone into minetest_game since 5.2.0 was released, it's mostly simple fixes and PRs prepared by other contributors.
Regarding "supporting", the options of shipping with no games / shipping with more games have been discussed recently. Shipping with none was deemed to be the better option but is not quite ready yet IMO until the mainmenu improves to better guide users towards installing games.
Perhaps it's time to formally put MTG into a bugfix only mode.

Zughy wrote:
Mon Jun 15, 2020 13:25
But let's remain on topic: there is, what, 4 of you? 5?
There are 11 coredevs of which 6 are long-term inactive, so that indeed leaves 4-5 who spend some of their free time on (primarily) the engine.
I'd say however that there are effectively only 3 devs that actively review engine PRs (which is the problem this topic draws attention to). This is nowhere near enough to cope with amount of contributions.
Mods: Mesecons | WorldEdit | Nuke & Minetest builds for Windows (32-bit & 64-bit)

User avatar
Hume2
Member
Posts: 631
Joined: Tue Jun 19, 2018 08:24
GitHub: Hume2
In-game: Hume2
Location: Czech Republic

Re: There are too many PRs

by Hume2 » Post

@Zughy, your words may be nice to listen but they aren't worth anything. If you want to move Minetest forwards, you should actually contribute and not only talk about how you would reorganise whole team.
If you lack the reality, go on a trip or find a job.

User avatar
Zughy
Member
Posts: 276
Joined: Thu Mar 26, 2020 18:23
GitHub: belongs_to_microsoft
In-game: Zughy
Location: Italy
Contact:

Re: There are too many PRs

by Zughy » Post

sfan5 wrote:
Tue Jun 16, 2020 12:45
But even if I wanted to, I cannot run engine development on my own
And you shouldn't, the system MT has got right now is perfect, because to err is human. I mean, in the PR erred 4 of us
sfan5 wrote:
Tue Jun 16, 2020 12:45
Shipping with none was deemed to be the better option but is not quite ready yet IMO until the mainmenu improves to better guide users towards installing games.
This is partially my bad, as I've been having a sketch of a main menu idea in my rucksack for weeks and I keep postponing it
sfan5 wrote:
Tue Jun 16, 2020 12:45
This is nowhere near enough to cope with amount of contributions.
I'd really like to help you, but I must ask: can we (community) expect anything of what discussed here (starting from the fist post) to be taken in consideration or things will just keep going as usual? In a way or another, we tried answering the "one million dollar question", but again the final decision is and it will always be yours

User avatar
Oil_boi
Member
Posts: 139
Joined: Mon Jan 28, 2019 13:12
GitHub: oilboi
IRC: oilboi
In-game: oilboi
Contact:

Re: There are too many PRs

by Oil_boi » Post

Damn I had 100 more prs I wanted to add >:(
I like to make mods :D
I also made a patreon!

CalebJ
Member
Posts: 339
Joined: Sat Dec 29, 2018 19:21
GitHub: KaylebJay
IRC: KaylebJay
In-game: CalebJ
Location: Tunnelers' Abyss

Re: There are too many PRs

by CalebJ » Post

Oil_boi wrote:
Tue Jun 16, 2020 14:44
Damn I had 100 more prs I wanted to add >:(
:D
If you want realism, you're in the wrong place. Get off your mobile.

User avatar
Lone_Wolf
Member
Posts: 2439
Joined: Sun Apr 09, 2017 05:50
GitHub: LoneWolfHT
IRC: Lone_Wolf or LoneWolfHT
In-game: Lone_Wolf
Location: Not there, THERE!

Re: There are too many PRs

by Lone_Wolf » Post

I said here viewtopic.php?f=7&t=24633: "If you review one of my PRs I will review one of yours. If you want a specific one reviewed link it here"

Even if your review doesn't count, if you find problems then the people who's review does count won't waste their review time just finding bugs.
My ContentDB -|- Working on Voxel Knights -|- Minetest Forums Dark Theme!! (You need it)

CalebJ
Member
Posts: 339
Joined: Sat Dec 29, 2018 19:21
GitHub: KaylebJay
IRC: KaylebJay
In-game: CalebJ
Location: Tunnelers' Abyss

Re: There are too many PRs

by CalebJ » Post

Lone_Wolf wrote:
Tue Jun 16, 2020 16:33
Even if your review doesn't count, if you find problems then the people who's review does count won't waste their review time just finding bugs.
Great idea LW! I should do this too. :)
If you want realism, you're in the wrong place. Get off your mobile.

User avatar
runs
Member
Posts: 2153
Joined: Sat Oct 27, 2018 08:32
GitHub: runsy

Re: There are too many PRs

by runs » Post

Why not Minetest in Google Summer Of Code?
🤑 Patreon |Minetest Chat | 👧 Samantha | ⛏️ Juanchi Game | 🐱 Petz

User avatar
Zughy
Member
Posts: 276
Joined: Thu Mar 26, 2020 18:23
GitHub: belongs_to_microsoft
In-game: Zughy
Location: Italy
Contact:

Re: There are too many PRs

by Zughy » Post

So, rant aside, my take about how to solve the problem:

- write a documentation of the engine, to both spur and help new members who'd like to contribute, AKA people who weren't here when Minetest was born (and simpler to understand). Ruben's modding book is extremely useful for making mods: it doesn't have to be that much extensive, even a head start can help a lot
- the previous point can also be the perfect time to clean some hypothetical messy code. Throwing some Marie Kondo at it will definitely help everyone
- this also mean slowing down the pace of MT development for a while but I think is necessary: because the more you wait and the more MT gets popular, the worse it'll get. A lot of people will appear out of the blue but, due to the first point, not a lot will be able to help you. A better main menu, entities API and whatnot will be nice at first, but in the end they'll just bring more problems if the amount or core devs stays the same while the amount of users keep growing. This can already happen with mods if some mod gets popular, there's no need to worsen the situation
- this is a bit radical but, especially because of the previous points: close issues and PRs once in a while if you feel overwhelmed. I don't think anybody will be offended. Unplugging every so often is nice
- please write down a few solid milestones. Not just about bugfixing, but something that could represent a roadmap. Where is Minetest going? As I stated before, at the end of the day you decide. We can provide you 500 PR and 1000 opinions, but the final decision is yours. Take your time, compare your opinions, debate. And only then do

In short: spring cleaning


About PRs, even if it's not that relevant if compared to what I suggested before:
- if something needs rebase and the author doesn't show up in a span of time of one month, the PR should be closed. It's up to the "pullers" caring about their work, not yours
- if the author does show up, but after another month doesn't commit anything useful, same as above

That will get rid of about 20 PRs

User avatar
Hume2
Member
Posts: 631
Joined: Tue Jun 19, 2018 08:24
GitHub: Hume2
In-game: Hume2
Location: Czech Republic

Re: There are too many PRs

by Hume2 » Post

runs wrote:
Tue Jun 16, 2020 20:14
Why not Minetest in Google Summer Of Code?
In 2016, it was said that the chance of a game being approved in GSOC (-ialism :D) is 1:7. I don't know how it is now but it's definitely not better. Also note that they want to know a lot of personal data and I don't know how many devs would like to offer them.
If you lack the reality, go on a trip or find a job.

User avatar
runs
Member
Posts: 2153
Joined: Sat Oct 27, 2018 08:32
GitHub: runsy

Re: There are too many PRs

by runs » Post

I think the problem is the lack of developers. Getting tougher and erasing PR will effectively lower the number of PRs, but it won't solve anything. What difference does it make if there are 1000 or 10? If the code's not going to be improved, LOL. Well, yes, the hypocrisy of the low numbers. :-D

Also closing or discarding PRs is dangerous, good code can be lost and contributors can be disappointed.

The problem I have already said is the lack of core developers. And it works:

1) Minetets being more popular. Which brings us to 2.
2) Getting more core developers.
3) A priority system and roadmaps to priorize PRs.
4) That the developers get along better, right now they don't support each other or collaborate with each other at all. From time to time, quarrels and rivalries arise. There are many personal differences. I meant as far as the Minetest vision is concerned. Now you don't see a team, but independent entities and that each one goes to its own.
🤑 Patreon |Minetest Chat | 👧 Samantha | ⛏️ Juanchi Game | 🐱 Petz

User avatar
Linuxdirk
Member
Posts: 2741
Joined: Wed Sep 17, 2014 11:21
In-game: Linuxdirk
Location: Germany
Contact:

Re: There are too many PRs

by Linuxdirk » Post

runs wrote:
Wed Jun 17, 2020 11:16
What difference does it make if there are 1000 or 10?
1000 for developers: Damn, too much work. Better spend my dev time on things I’m interested in instead of working on those PRs, I can’t bring the number down significantly so I don’t try to bring it down at all because it frustrates me.

1000 for contributors: Wow, 1000 PRs, they seem not to care about PRs so I guess I just fix the code for myself or just file a PR and do not care anymore because obviously no-one works on the PRs.

10 for developers: Just 10 PRs, this can be done quite easily. Everyone of us takes one PR to implement or decide on not implementing it. If I take just one of them I lowered the amount of PRs by 10%, this feels good and I like working like this.

10 for contributors: Wow, just 10 PRs, they know how to handle stuff and I’ll happily file my PR and actively help implementing it because 10 PRs are not that much, so devs will check my PR very soon.

User avatar
runs
Member
Posts: 2153
Joined: Sat Oct 27, 2018 08:32
GitHub: runsy

Re: There are too many PRs

by runs » Post

Linuxdirk wrote:
Wed Jun 17, 2020 12:04
runs wrote:
Wed Jun 17, 2020 11:16
What difference does it make if there are 1000 or 10?
1000 for developers: Damn, too much work. Better spend my dev time on things I’m interested in instead of working on those PRs, I can’t bring the number down significantly so I don’t try to bring it down at all because it frustrates me.

1000 for contributors: Wow, 1000 PRs, they seem not to care about PRs so I guess I just fix the code for myself or just file a PR and do not care anymore because obviously no-one works on the PRs.

10 for developers: Just 10 PRs, this can be done quite easily. Everyone of us takes one PR to implement or decide on not implementing it. If I take just one of them I lowered the amount of PRs by 10%, this feels good and I like working like this.

10 for contributors: Wow, just 10 PRs, they know how to handle stuff and I’ll happily file my PR and actively help implementing it because 10 PRs are not that much, so devs will check my PR very soon.
I understand that, but my vision is different. I think that "There are too many PRs" is not a problem at all.

- It's better 1,000,000 PR than 1,000 because it means:

1) People are willing to contribute.
2) Obviously there will be PR of various types: good, bad, mediocre, etc. but the greater the amount the better the probability of better PR = Better Minetest.

The problem, I think, is the lack of core developers who can manage them. It wouldn't be good for me because of that:

- Limit PRs.
- Close PRs because they can't be handled.

It would be good, medicine to solve it and maintain a lot of good PRs:

- Increase the number of core developers.
- Core developers will always be able to choose the ones they like best.

Anyway contributors always think that their PRs are fundamental and would like them to be implemented, otherwise they would not publish them to github :-)
🤑 Patreon |Minetest Chat | 👧 Samantha | ⛏️ Juanchi Game | 🐱 Petz

User avatar
Zughy
Member
Posts: 276
Joined: Thu Mar 26, 2020 18:23
GitHub: belongs_to_microsoft
In-game: Zughy
Location: Italy
Contact:

Re: There are too many PRs

by Zughy » Post

runs wrote:
Wed Jun 17, 2020 11:16
4) That the developers get along better, right now they don't support each other or collaborate with each other at all. From time to time, quarrels and rivalries arise. There are many personal differences. I meant as far as the Minetest vision is concerned. Now you don't see a team, but independent entities and that each one goes to its own.
Fixed
Image
(who is who is completely random. RIP Phoebe)

Also, agree with Linuxdirk

User avatar
Wuzzy
Member
Posts: 4142
Joined: Mon Sep 24, 2012 15:01
GitHub: Wuzzy2
IRC: Wuzzy
In-game: Wuzzy

Re: There are too many PRs

by Wuzzy » Post

There has been a lot of talk, but the problem has not gone away … at all.
Minetest has been consistently been over 150 OPEN PRs for months now, and there is no sign that it will ever go below 150 again.
Whatever is done to reduce the number, it's not working.

I feel that there's still not much organization regaring PRs, it feels rather chaotic to me. I suggest adding labels to do a small pre-review for all PRs. A pre-review is only a very superficial look at the PR, in order to classify it and to decide how important it is.

There are the labels that can be used after a pre-review:
- “acceptable”: This is given after a superficial review of the concept/idea of a PR alone. It is given when there is no major dispute, the concept aligns with the Minetest development philosophy, and the PR adds value without clashing with existing concepts
- “disruptive”: Anything that changes a major as. This label should NOT be meant as an automatic death sentence, it's just a warning. Sometimes, being disruptive is the only way to move forward. An example for a disruptive change in the past was the removal of the Sneak Glitch
- “complexity: high/med/low”: Rough estimate to show how “difficult” it is to review and merge the PR. As a rule of thumb, the more files and code you touch, the more complex.
- high complexity: Introduces a major feature, touches many different files in many unrelated locations at once, touches networking stuff, chance of attracting conflicts is high
- med complexity: Touches a couple of files/code lines, but it's mostly in the same area of the codebase, so a conflict only arises if a competing PR touches the same area
- low complexity: Only touches a handful of files/code or even just text files. Conflicts are either unlikely or easy to resolve
- “pre-review pending”: Given automatically to all new PRs. This label means what it says: The pre-review is not done. Remove this label when acceptability, disruptivity and complexity have been decided on.

More strategy:
ASSUMING, your time as PR reviewer is limited, I propose several review strategies to (hopefully) make the review process for PRs more efficient without requiring additional review time, but by changing what to do during the review time.

1) The MOST important thing (seriously): Close ancient PRs that aren't going anywhere. Specifically, PRs that are: >1 years old, have an unresponsive author and have collected conflicts like moths should be closed.
2) High-complexity PRs should be reviewed before low-complexity ones.
3) PRs that are highly controversial should be quickly closed and put “on hold”. Maybe add a special label just for that. Then, an issue should be opened to replace the PR, to discuss design FIRST. When there is a consensus reached on a given design, PRs can then shine again.

Details and rationales about 1-3:
1) There are sadly a TON of PRs that fit that description RIGHT NOW. I suppose this single step will already bring down the PR count massively. Let's be real: It's unlikely these PR will become active again, so it's foolish to keep them open infinitely in the hopes the author might someday magically reappear. Just close them, but add the label “PR grew too old”. This signifies the PR as idea might still be acceptable, but it has failed technically. There's also still a (small) chance someone might try to ressurrect the PR … But now comes the radical and evil part: Even do this when the PR's failure is 100% not the author's fault, the PR got 1 or even 2 approvals, the author followed every single request, and the only reason the PR is stalling because of lack of reviews. Why? Because the large amount of conflicts means the PR has already failed, and needs probably to be rewritten anyway. Yes, this will be deeply frustrating to everyone, but the PR is a lost cause anyway, so keeping it open forever is just lying to ourselves. The damage has already been done a long time ago, it's only about admitting it.
2) Why? Because these are the most likely to attract conflicts quickly (since many files are touched), and this can get ugly very soon. The faster they are dealt with, the less likely you will have to deal with complexity. Low-complexity PRs, on the other hand, can be dealt with later, as the chance of conflicts is much lower.
3) Why? Because if a PR is controversial is a sign that there is disagreement about the design of some feature, and we should use issues first to discuss design principles, NOT PRs. It's better for everyone when we nailed down the design FIRST, and then move to implementation, not vice-versa. But only reserve this step for major community splits, not for every single disagreement.
My creations. I gladly accept bitcoins: 17fsUywHxeMHKG41UFfu34F1rAxZcrVoqH

User avatar
Zughy
Member
Posts: 276
Joined: Thu Mar 26, 2020 18:23
GitHub: belongs_to_microsoft
In-game: Zughy
Location: Italy
Contact:

Re: There are too many PRs

by Zughy » Post

I don't agree, it sounds highly messy

User avatar
Lone_Wolf
Member
Posts: 2439
Joined: Sun Apr 09, 2017 05:50
GitHub: LoneWolfHT
IRC: Lone_Wolf or LoneWolfHT
In-game: Lone_Wolf
Location: Not there, THERE!

Re: There are too many PRs

by Lone_Wolf » Post

I'd say try it and see if it actually is messy
My ContentDB -|- Working on Voxel Knights -|- Minetest Forums Dark Theme!! (You need it)

User avatar
Zughy
Member
Posts: 276
Joined: Thu Mar 26, 2020 18:23
GitHub: belongs_to_microsoft
In-game: Zughy
Location: Italy
Contact:

Re: There are too many PRs

by Zughy » Post

Lone_Wolf wrote:
Wed Jun 17, 2020 16:54
I'd say try it and see if it actually is messy
the point of this is core devs not having enough dev time. If you bring more bureaucracy, it won't do much good. It's basically the summary of this

Post Reply

Who is online

Users browsing this forum: No registered users and 4 guests