There are too many PRs

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

rubenwardy wrote:
Tue Jun 23, 2020 19:24
ANAND / ClobberXD is now a core developer. More people are being considered
Congratulations to him. A great addition to the team.
If you want realism, you're in the wrong place. Get off your mobile.

User avatar
Linuxdirk
Member
Posts: 2743
Joined: Wed Sep 17, 2014 11:21
In-game: Linuxdirk
Location: Germany
Contact:

Re: There are too many PRs

by Linuxdirk » Post

rubenwardy wrote:
Tue Jun 23, 2020 20:49
The wiki is the official source for engine development guidelines and documentation, and always has been.
So there is no offline guidelines and documentation for the engine?

User avatar
rubenwardy
Moderator
Posts: 6431
Joined: Tue Jun 12, 2012 18:11
GitHub: rubenwardy
IRC: rubenwardy
In-game: rubenwardy
Location: United Kingdom
Contact:

Re: There are too many PRs

by rubenwardy » Post

There is not, other than the code itself and some txt files in the repo

User avatar
LMD
Member
Posts: 967
Joined: Sat Apr 08, 2017 08:16
GitHub: appgurueu
IRC: appguru[eu]
In-game: LMD
Location: Germany
Contact:

Re: There are too many PRs

by LMD » Post

rubenwardy wrote:
Tue Jun 23, 2020 19:24
ANAND / ClobberXD is now a core developer. More people are being considered
Great!
My stuff: Projects - Mods - Website

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

Wait what? I always thought the engine parts of the Dev Wiki are official. Or at least semi-official. Since they contain development guidelines and stuff.

Anyway, let me stress again that the Dev Wiki IS WRONG on most Lua API claims. Most of the Dev Wiki just ripped stuff from lua_api.txt years ago, which creates a ton of pointless redundancy and inconsistency. Those redundant pages should be deleted. ALL OF THEM.

Fun fact: I also was asked to become a core dev one day. But I don't think I'm competent or willing enough to actually do it.
My creations. I gladly accept bitcoins: 17fsUywHxeMHKG41UFfu34F1rAxZcrVoqH

User avatar
jas
Member
Posts: 593
Joined: Mon Jul 24, 2017 18:15
IRC: Freenode
Location: IRC

Re: There are too many PRs

by jas » Post

Just delete it (the dev wiki) until someone makes a script that automatically generates it based off lua_api.txt. I think someone else already did that, so just swap 'em out?

Wuzzy thank you for being honest with yourself, but for what it's worth I believe in your abilities. When you're not feeling like it, and have little inclination, that changes everything of course.

Fun fact: Wuzzy railed for YEARS to remove sneakjumping

Termos
Member
Posts: 400
Joined: Sun Dec 16, 2018 12:50

Re: There are too many PRs

by Termos » Post

Wuzzy wrote:
Thu Apr 30, 2020 15:56
The Minetest development procedure has a huge problem: There are too many open PRs (pull requests).
IMO the root problem is organizational structure, or lack thereof. Everybody's equal, and that's cute and all, but that also means many conflicting opinions and visions (or visionlessnesses) and random outcomes. Without a strong decision center having a clear idea of where they want to get this project, any ensuing controversies tend to deadlock or end in catastrophe like the infamous sneak bugs which FUd the development of roughly half the engine. This project's lost its head and hasn't grown a new one, so the staff as a whole can't help but function like a decapitated chicken regardless of individual members' qualities.

In this situation I'm not sure recruiting more staff is going to help much, it has the potential to create more chaos and more opportunity to talk PRs to death.

User avatar
rubenwardy
Moderator
Posts: 6431
Joined: Tue Jun 12, 2012 18:11
GitHub: rubenwardy
IRC: rubenwardy
In-game: rubenwardy
Location: United Kingdom
Contact:

Re: There are too many PRs

by rubenwardy » Post

The vast majority of pull requests are uncontroversial

The sneak bug issue only lasted a week until a resolution was merged. If you're referring to the presence of old vs new move, this was only added after many community members complained about the sneak bug being fixed. Old move is a way to keep those members happy without compromising the new move code

User avatar
Linuxdirk
Member
Posts: 2743
Joined: Wed Sep 17, 2014 11:21
In-game: Linuxdirk
Location: Germany
Contact:

Re: There are too many PRs

by Linuxdirk » Post

rubenwardy wrote:
Wed Jun 24, 2020 13:29
The vast majority of pull requests are uncontroversial
If the vast majority of PRs in uncontroversial, then why are there so many open PRs? Either reject or merge.

User avatar
rubenwardy
Moderator
Posts: 6431
Joined: Tue Jun 12, 2012 18:11
GitHub: rubenwardy
IRC: rubenwardy
In-game: rubenwardy
Location: United Kingdom
Contact:

Re: There are too many PRs

by rubenwardy » Post

Because they're all valid, reviewing is a time consuming and difficult process

If we just merged, then we'd get complaints about bugs and instability. A lot of Minetest's code problems were caused by just merging PRs

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

Linuxdirk wrote:
Wed Jun 24, 2020 13:46
then why are there so many open PRs?
I think this has been answered/guessed at multiple times in this thread
My ContentDB -|- Working on Voxel Knights -|- Minetest Forums Dark Theme!! (You need it)

User avatar
Linuxdirk
Member
Posts: 2743
Joined: Wed Sep 17, 2014 11:21
In-game: Linuxdirk
Location: Germany
Contact:

Re: There are too many PRs

by Linuxdirk » Post

rubenwardy wrote:
Wed Jun 24, 2020 13:48
If we just merged, then we'd get complaints about bugs and instability.
And how is this different from the current situation?

User avatar
rubenwardy
Moderator
Posts: 6431
Joined: Tue Jun 12, 2012 18:11
GitHub: rubenwardy
IRC: rubenwardy
In-game: rubenwardy
Location: United Kingdom
Contact:

Re: There are too many PRs

by rubenwardy » Post

I think you're unable to see how bad it would get. Imagine Freeminer, but worse

User avatar
Linuxdirk
Member
Posts: 2743
Joined: Wed Sep 17, 2014 11:21
In-game: Linuxdirk
Location: Germany
Contact:

Re: There are too many PRs

by Linuxdirk » Post

There must be something between having nearly 1000 issues and more than 150 PRs on one side and merging everything without testing on the other side.

User avatar
Lejo
Member
Posts: 715
Joined: Mon Oct 19, 2015 16:32
GitHub: Lejo1
In-game: Lejo

Re: There are too many PRs

by Lejo » Post

As Wuzzy already said there should be a fast and clear "review plan".
There should be a fix order to not loss any time on doing steps for nothing
Example order:
  1. Is the goal accepted and clearly discussed? (Maybe only allow PRs with previous Issue)
  2. Does the PR solves this Problem?
  3. Were builds successful?
  4. Is the Codestyle correct?
  5. Does the PR work?
We don't need to review a PR with a not accepted goal...
What about adding idea approvals to Issues so that a PR changing that is more trival?

This plan should also be visible to everybody so that you can do you PRs easy to review and more likely to get merged

User avatar
paramat
Developer
Posts: 3700
Joined: Sun Oct 28, 2012 00:05
GitHub: paramat
IRC: paramat
Location: UK

Re: There are too many PRs

by paramat » Post

Astrobe wrote:
Sat Jun 20, 2020 09:22
I think the Pull Requests feature popularized by Git gave to some the wrong idea.

PRs actually consists in hundreds of people |i]pushing[/i] things on a dozen of people (at best). No wonder PRs eventually overflow. Just looking at them takes too much time already. It's like a brain DDOS. That's the bad side of "social coding".
This is actually a very good point, and something i think about.

Becoming overwhelmed by contributions and requests is inherent when you have open source software that is given away free.
Anyone on the internet (4.5 billion people) can ask for features or contribute code.
But because core devs are unpaid, very few people can find time to be a core dev, and those that can have very limited time and are often already exhausted.

This would not be a problem if it was either:
Closed source with a free game and unpaid devs.
Or,
Open source with a priced game and paid devs.

Yet again, as with so many discussions about Minetest, it mostly comes back to this:
"It is a free game, what do you expect?"
"If you want improved development, pay the core devs full time wages"
=D
Many people have expectations that are too high and not much different to expectations of commercial closed source games.

User avatar
paramat
Developer
Posts: 3700
Joined: Sun Oct 28, 2012 00:05
GitHub: paramat
IRC: paramat
Location: UK

Re: There are too many PRs

by paramat » Post

Lejo wrote:
Tue Jun 23, 2020 19:23
Another thing about dev-time.
I think it would be useful to split lua and core development up.
C++ Minetest Developement is a lot more harder to understand than just modding.
Actually no, the Lua part of the engine (which is not 'modding') is just as difficult to work with as the C++ parts.
Anyone working on the Lua parts of the engine still needs to be equally as trustworthy and still needs the same experience and knowledge, so the requirements to become a core dev are no lower.

User avatar
paramat
Developer
Posts: 3700
Joined: Sun Oct 28, 2012 00:05
GitHub: paramat
IRC: paramat
Location: UK

Re: There are too many PRs

by paramat » Post

Wuzzy wrote:
Wed Jun 17, 2020 15:43
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.
This already happens. We recently reduced the time period for neglected PRs to 6 months. There is also this existing dev rule:
"If a pull request or an issue does not get a response from its author in one month (when requiring more details), it is closed."
Wuzzy wrote:
Wed Jun 17, 2020 15:43
2) High-complexity PRs should be reviewed before low-complexity ones.
I disagree, because simple PRs are just as important and valuable as complex PRs.
Also, of course, because we do not have enough time to review all PRs, this rule would mean all simple PRs are never reviewed or merged. If anything decides what PRs are reviewed and merged it should be something like 'benefit and suitability'.
Complex PRs are a headache to review and test, mixing these with simple PRs is essential for core dev sanity.

User avatar
paramat
Developer
Posts: 3700
Joined: Sun Oct 28, 2012 00:05
GitHub: paramat
IRC: paramat
Location: UK

Re: There are too many PRs

by paramat » Post

I wrote:
> All these altready happen, we close everything we possibly can, and after 6 months of no activity.
Wuzzy wrote:
Wed May 13, 2020 14:48
The data suggests otherwise. The oldest open PR is from 2016 and there are close to 100 PRs that are at least 6 months old.
No, the data does not suggest otherwise =) After closing every PR neglected (changes required but not done, or no response to an important enquiry) for more than 6 months, what you see is simply what is left. All the 150 open PRs are not neglected and are only delayed due to lack of core dev time.
If a PR is good in every way, rebased, but is still unreviewed after 4 years due to lack of core dev time, there is not really a reason to close it while keeping newer PRs open.

User avatar
Linuxdirk
Member
Posts: 2743
Joined: Wed Sep 17, 2014 11:21
In-game: Linuxdirk
Location: Germany
Contact:

Re: There are too many PRs

by Linuxdirk » Post

paramat wrote:
Wed Jun 24, 2020 22:25
If a PR is good in every way, rebased, but is still unreviewed after 4 years due to lack of core dev time, there is …
… something really wrong in handling PRs, yes.

User avatar
pandaro
Member
Posts: 323
Joined: Sun Jan 08, 2012 21:34
GitHub: pandaro
Location: behind

Re: There are too many PRs

by pandaro » Post

i have the definitive solution:
delete all the PR and the issues -> reset everything.

User avatar
jas
Member
Posts: 593
Joined: Mon Jul 24, 2017 18:15
IRC: Freenode
Location: IRC

Re: There are too many PRs

by jas » Post

I'm using the mouse bind PR

I know deleting the PR doesn't delete the original repo/branch but, I'm still using 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

But because core devs are unpaid, very few people can find time to be a core dev, and those that can have very limited time and are often already exhausted.
Oh c'mon, that's plain nonsense: you don't become core dev of projects like MT for the money. You probably do it because of a moral compass (FOSS and such) or because you simply enjoy it. And, speaking personally, the former gives a pretty good boost and a sense of resilience
paramat wrote:
Wed Jun 24, 2020 21:11
Yet again, as with so many discussions about Minetest, it mostly comes back to this:
"It is a free game, what do you expect?"
"If you want improved development, pay the core devs full time wages"
=D
Many people have expectations that are too high and not much different to expectations of commercial closed source games.
Say that to Blender, Krita or Inkscape. They're basically no different than the closed source standard now. You're collaborating with the CERN for fork sake, I mean, the CERN! That's simply mindblowing, guys! So yes, I expect a lot from a project like this.
If a PR is good in every way, rebased, but is still unreviewed after 4 years due to lack of core dev time, there is not really a reason to close it while keeping newer PRs open
Yeah, 4 people in 4 years didn't have a moment to review a PR. See, that's what I call offensive and ridiculous

User avatar
rubenwardy
Moderator
Posts: 6431
Joined: Tue Jun 12, 2012 18:11
GitHub: rubenwardy
IRC: rubenwardy
In-game: rubenwardy
Location: United Kingdom
Contact:

Re: There are too many PRs

by rubenwardy » Post

Zughy wrote:
Thu Jun 25, 2020 23:44
Oh c'mon, that's plain nonsense: you don't become core dev of projects like MT for the money.
Money = time. A paid developer is able to work full-time
Zughy wrote:
Thu Jun 25, 2020 23:44
Yeah, 4 people in 4 years didn't have a moment to review a PR. See, that's what I call offensive and ridiculous
We've reviewed many many hundreds of PRs in 4 years

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

rubenwardy wrote:
Fri Jun 26, 2020 00:01
Money = time. A paid developer is able to work full-time
Sure, but how someone can tell how much money Minetest earns? They can't, or better said:
  • Paramat hid the number (but in the end I guess it's ok, it's his personal profile);
  • I can't find any sfan5 nor Krock funding link, so I couldn't support them even if I wanted to;
  • There is no Minetest Patreon/Liberapay/KoFi whatsoever. There actually is a "donate" section on the official site, but according to what it says, those money directly end up to celeron55 ("You can tip money to celeron55, to be mostly used for hosting of websites and overall maintenance effort on the project").
How about creating a Minetest profile on Liberapay, for example? (I'm not suggesting Patreon because of their damn high fees, nor KoFi for their privacy policy)
rubenwardy wrote:
Fri Jun 26, 2020 00:01
Zughy wrote:
Thu Jun 25, 2020 23:44
Yeah, 4 people in 4 years didn't have a moment to review a PR. See, that's what I call offensive and ridiculous
We've reviewed many many hundreds of PRs in 4 years
I'm no doubting that, but paramat was talking about a specific PR (I guess this one?)

Post Reply

Who is online

Users browsing this forum: No registered users and 7 guests