pouët.net

Oh and I don't really blame Gargaj for the new code not being up to par 2013. At the time (five years ago) it was propably the best idea to do what he did. But it shouldve been deployed a long time ago already since today, it's really old and crappy as it is. Deploying another old hard coded php site just doesnt make sense as long as the old one works since the new features are not in the hard coded modules but rather in the api which should be in the bottom layer of the code base not built on top of php-whackiness.

And yes, the new datamodel might be alot better so it might be a nice approach to use the migration code to a new datamodel and work from there but I don't generally like this approach since its extremly hard to validate the data integrity of migration scripts. But if it works it might be a good start but I don't see why you should deploy a code base that havent been sharp and probably will be full of new minor or major bugs after deployment if you want to go for the more serious approach.

Gargaj: I was just trying to get a license-clarification so changes could be incorporated into the live code-base even if you decide not to take part in further development. Without a license, legally speaking it's still not quite open source :/

what kusma said.

Would public domain work? Or is there one that's even less hassle for everyone involved?

that would do

What lator(!) said. No, really. Maybe hell just froze over, but for some reason he's mostly right. :P

v2 has problems, not just the ones mentioned several times, like the lack of separation of markup, SQL-stuff and PHP script. It's also the influences of oldschool desktop game/demo programming that seem strange and out of place (no, really, you _don't_ need all those render methods). Also, some pet peeves like the onset of file count inflation (thank you, Java) and whatnot. Maybe that's just me. But still, DB, API and the website as a client on top of it seems to be a current and reasonable approach. This would be more like v2.5 though. :)

Gargaj: http://de.wikipedia.org/wiki/WTFPL. PD should be fine though.

Tomaes: This just in, not everyone speaks german!
(and I'm sure Gargaj knows this license anyway)

gargaj: Public domain is plenty good, yes. I take that's your answer, then?

Sigh, okay, added an unlicense to the ZIP to keep me from issuing a C&D to whatever authority happens to be over Pouet.

Tomaes: Have you tried recreating the Pouet layout (and I mean as close as I managed) with using a templating engine without it becoming extremely cumbersome AND fast? (Granted, at the time Smarty was the best, I'm not sure if someone made better since, but it was both sluggish and a disaster to maintain - no two pages look the same and you win nothing.) And yeah, the ORM solution I added doesn't have CRUD but really, it nearly never needs it either, on the other hand the relation handling in it is automated and works better than any of the solutions I tried (and dfox knows that I kept going through a few).

although gargaj's code is out, still no repository around...

I am sure if analogue find the code useful he will merge / update current code base and set up a repository for it. If noone else does it.

I guess the main question is if its time-wise worth it depending on the actual road map for a new pouet api most people (not gargajs fan club) agree upon is the proper way to go for the long run.

I'd rather see analogue created a draft for a api road map and and the project was open from the very beginning.

btw said fanclub is available at https://www.facebook.com/pages/Gargaj/196764203668940

like traded for source code ;)

API is fine and good but the rest of the buzzwords that Analogue wrote just make me cringe.

MVC -> Yeah, didn't quite work out in the last 20 years, and in most web projects it's just used to make two php files out of one and increase complexity without any measurable advantage. Really. Name ONE example where separation of view and controller actually _did_ something. API/webpage is separation enough, really. Also, php's already hobbirble code readability goes further down the drain when you need to actually know the framework to find out in which file a certain app-level functionality might hide.
ORM -> "You shouldn't have used SQL anyway".
REST -> sounds totally reasonable in theory, we already have http, why don't we use it as intended, yaddayadda. But as soon as you use it (and I've collected lots of experience implementing and using restful services in the last year) you'll find out the only thing it does is inherit all the problems http has without giving anything back in return. Also, everything but the most trivial queries (> "select with where only") will break the paradigm anyway. Having GETs with easy path names in the URL is ok, all the rest (haha) is bullshit. Eg. JSON-RPC over http is simple enough without sacrificing any flexibility.

Also the page should really mostly work with Javascript disabled. There, I said it. Because there's more than PCs in the demoscene. Oh, and what Knos said about snappiness. I can't stand the seconds "modern" pages take to swap in all their overhead either.

Awesome, took 17 pages but it was worth it I guess.
I wouldn't let mog own your repo, but I guess I have a bad opinion about him because of this thread.

Repo for 1.0 created and public, I'll push the 1st fixes tonight.
https://github.com/lra/pouet.net
Especially now that I know that the migration is not that great, no offense G.

Let's see what happens from now on.

Analogue: you should probably use a different metric for success than a low (?) thread page count.

mog: Hear hear =)

Gloom: I certainly would not have used the word success. A lot of trust have been lost on both sides.

kb: but, but... without javascript, how can we do beautiful things like this? :(

Because I've had it with this thread, the petty bickering (from both "sides" and that you guys even formed them at all), and the fact that you guys managed to ruin the only thing that was actually moving and getting anything done:


From your new repo is not legally sound. You cannot just assign GPL to all that, as you have noted that you had contributors to the code other than you and it's quite likely they didn't sign a contributor agreement. So unless they're willing to give up their own implicit copyrights on their own contributions, assign them to you, or you're willing to rewrite those parts*, you can't just unilaterally relicense it. Even if you were the originator of the project.

* This might be tough without something better than a change log to identify the code in question.

Lessons learned from this fun entry in the annals of Pouet.net: when people fight, don't try to make peace. Let them scream at each other until they come to a conclusion.

With Flash, of course.