FOSS gamedev and creative worlds

The total lack of quality control and a truly stable release is totally frustrating.

My server have a pretty large bandwidth, about 100 mbps. I have a good band too, 10 mbps. That means I can download things at 1.2 MB/s (true and actual real life speeds).

My mods folder have 19 MB, but it takes >5 minutes to get all media and item textures when it should take 19 seconds at most!


A lot of buggy commits are constantly pushed to GIT and made available to download. That's unacceptable.


I had tons of game servers in my life already, even on dial up connections.

Don't get me wrong, minetest core developers are brilliant people who really know to code well and are fast working to resolve issues but it seems like there is no one controlling that, testing and doing quality control.
It also seems like they care more about implementing new features all the time than actually fixing bugs and optimizing the code.

As you can read in my server's site and topic, I'm shutting now my server for now.


My suggestion is.

• Make a board that decides what features to implement.
• Make a testing team, to test new features before they are made public.
• Do more bug hunt.
• Optimize speed. Your protocol is not fast!

Sorry for that, I'm just frustrated and disappointed.

Sounds very much like the problems we're experiencing at VanessaEs survival server. The extra media data for the mobf animals seems to have been too much :-(. Hope that bug can be identified and then eliminated.

As far as I understand the development process, it already takes two core devs to approve of a pull request, and in theory it ought to be tested. Lots of complaints are coming from people who feel that their requests are not handled fast enough (it seems to take a lot of time sometimes). In practice, one particular area might be of intrest to one core dev, while the others either do not care particulary about that topic or have no clue about it, so finding testers may not be easy.

I'm sure the core devs will approve of helpful bug reports and coding help. Most do not really play and thus will not encounter some of the problems server operators encounter.

Megaf wrote:My server have a pretty large bandwidth, about 100 mbps. I have a good band too, 10 mbps. That means I can download things at 1.2 MB/s (true and actual real life speeds).
My mods folder have 19 MB, but it takes >5 minutes to get all media and item textures when it should take 19 seconds at most!

Do you use remote_media? If not, that's from where the problem comes from: from-server downloads are not fast, to not eat all the server's bandwidth.

By the way, the lag that has been around large Minetest servers is present since 0.4.3 is released (or even a bit earlier)... I even experienced it back in 0.4.3.

http://freeminer.org/

has more optimization, and more features.

Yeah I think maybe a "suggestions" sub-forum here would be great.

I also think that bugs put in bugs and glitches need to be actually solved.
Most of the time someone answers like this "oh you need to copy and paste this bug fix to blabla".
It would be better if a developer answers saying: "thank you for reporting. This is now fixed."

CraigyDavi wrote:Yeah I think maybe a "suggestions" sub-forum here would be great.

I also think that bugs put in bugs and glitches need to be actually solved.
Most of the time someone answers like this "oh you need to copy and paste this bug fix to blabla".
It would be better if a developer answers saying: "thank you for reporting. This is now fixed."

Features Discussion is for suggestions, and modding general > mod requests for mod suggestions.

Things are a little better with remote_media options, yet ideal.

I still have some lag, but perhaps now is a local problem.

Still, work need to be done on client and server side when connecting to server, more things could be downloaded to client and kept there.

My server is back online thanks to sfan5, VannesaE and others.

Just had an idea. Could it be possible to hash the media files, send the hash to the client, and when the client has the same media it loads the files from the own minetest instance?

Actually everyone who uses a dev version of Minetest is part of our quality control.

PilzAdam wrote:Actually everyone who uses a dev version of Minetest is part of our quality control.

Erthome has a fair amount of experience in software development, so he should be of help here.

Casimir wrote:Just had an idea. Could it be possible to hash the media files, send the hash to the client, and when the client has the same media it loads the files from the own minetest instance?

Guess what the client does since 0.4.3

I am much too new here to even consider helping core developers directly, and frankly I am not much good at doctoring existing code...my guess is that nearly everyone already involved is probably better than I would be in that area...if I have anything to contribute in the development arena it will likely be of the quality control, process design variety...and that tends to be an area where people with years invested are least open to input from johnny-come-lately folks like me...

I'm here to play, offer useful ideas when I think I have them, give feedback on discussions in progress...

(To be perfectly honest, I have been planning a project for years which amounts to using a macro assembler to produce a cross platform assembly-code based high performance thin-interface client, and hybrid distributed server application environment to interface with a family of proprietary, firmware-based applications support modules, ultimately designed to run in concert with a state-logic based RDBMS and universal transaction model library and custom OS that I designed then prototyped using old 1MH ASIC technology in 1990. Blah blah blah...but among my reasons for being here, I am always on the lookout for people who have done hard time in such places as unruly open source environments, GET the point of public owned software, can put up with me, and want to be part of an industry revolution LOL but THAT is largely off-topic...)

SO, apart from all that, I am here to learn, and occasionally share what may seem like off the wall input because it isn't likely to come from elsewhere and might actually be interesting or useful for other folks from time to time.

One quick subject I'd like to throw in a couple cents on, based on some of the disconcert that I've seeing in others posts, we might want to start/resume distinguishing between end-users/server operators and beta-testers. It is really cool to have a short dev to user cycle when the users are all co-developers and testers, but that is an entirely different matter from trying to grow the player community...AND if everyone becomes a developer then the player base will soon become very scant. What's more, if there is any interest at all in nurturing a more immersive evolution of the product to include RPG or other storyline and theming capability into the public-facing server realm the server operators NEED some degree of backward world and client compatibility.

I don't mean this in a mean way, BUT the current environment here is very insular and almost secret clubbish...a new-comer is not going to be intimately familiar with the code or culture for a good while--the publicly available overviews of the code and other documentation are woefully un-useful, and as it stands I believe we are scaring off a lot of would be contributors...I have been mostly quietly but closely watching the community at work for a good 8 months now, and daresay I have a LOT of in game time for a relative newcomer...I have brought a dozen IRL friends and relatives to the servers and watched their reactions closely.

This all leads me to believe that the single greatest asset in this community is the community itself, even more so than the game--though without the game the community would vaporize, so they are symbiotic, and need to be recognized and provided for in that vein and light.

sfan5 wrote:Guess what the client does since 0.4.3

Oh, cool.

Erthome wrote:I am much... huge text ...and light.

I'm a ADHD person and could not read everything you wrote. Google luck with the ASIC thing, sounds really interesting.

Welcome do Minetest and you could join our IRC cannel #Minetest at irc://chat.freenode.net/
There you could talk directly to core devs. Your ideas surely could help, take a good look at the code, like connections.cpp and other stuff related to some protocols, your opinion may be of great value to us.

There already some folks forking minetest and changing some stuff to make it less laggy and more responsive, but we do need that changes to come back to minetest.

I don't think 0.5.0 could ever come soon enough.