00:16 < needs1> heinrich5991: So I implemented url format you suggested and it looks much better than what I had before
00:16 < needs1> heinrich5991: However, I could even more simplify it:
00:17 < needs1> By default players are sorted by rank, so I could make /players?p=1 be the same as /players/by-rank?p=1
00:17 < needs1> But now I have two urls pointing to the same resource
00:17 < needs1> Would you still go for that?
00:19 <@minus> implementing pages like that sucks
00:19 <@minus> if your information changes often
00:22 < needs1> minus: So you prefer not having such shortcut and stick to the full url /players/by-rank?p=1
00:22 < needs1> ?
00:23 <@minus> no, i mean using page numbers is/can be bad
00:23 < needs1> Then, how would you display just a part of the full player list?  I wan't really send the whole player list, it would be too big
00:24 < needs1> *I can't
00:24 <@minus> yeah, i have no solution
00:24 <@minus> ccan't even manage to find any rants on it
00:25 < needs1> It's seems common to use page number as url parameters
00:25 < needs1> Google does it
00:26 <@minus> well, it usually works well enough
00:27 < needs1> minus: What do you find wrong with page numbers?
00:27 < BotoX> pagination is a pretty standard thing
00:27 <@minus> the information changes
00:27 <@minus> maybe with ranks it's not wrong
00:31 < needs1> You mean, the page is not static?
00:37 <@minus> when you go from page 1 to page 2 you expect to get the info from where page 1 left off. but if the data changes meanwhile you may miss something
00:43 < needs1> I see, it's indeed an issue.  I've already tough about it when implementing the first teerank version and came to the conclusion that maintening old database entries just for the sake of consistency between pages is not worth the complexity it will bring.
00:45 < needs1> Well, I could have overestimated the complexity of such feature tho
00:46 <@minus> shouldn't be too difficult but may litter your database significantly
00:46 < needs1> It's indeed nice to have immutable page, now that I think of it
00:50 < needs1> Actually a list of player is about 3.2 for 100k players, let's say it is now 4 time bigger with the new version I'm about to release, if I do one save per day, it means 3.2 * 4 * 365 = 4.6GB per years
00:50 < needs1> *3.2MB
00:51 < needs1> Ah I need to multiply it by 3, because I also have a list of servers and clans now, so 14GB per years
00:51 <@minus> that's like
00:51 <@minus> nothing
00:52 < needs1> I have actually 2TB available on teerank server
00:52 < needs1> So 142 years to fill it
00:52 <@minus> if you cache  the latest 3 snapshots you'll not even have a performance problem with more data
00:53 <@minus> well, except if you don't have indices and run a full table scan each time \o/
00:53 < needs1> Actually it's a file based database so performance will not be a problem: I just have to put  snapshots in different directories or whatever
00:55 < needs1> In fact it's still a bit more complex than just storing the list of players infos: I already have a full historic of rank and elo values for each players...
00:55 < needs1> well whatever
00:55 < needs1> That's something I might endup doing
00:56 < needs1> But not for the current version for sure, too much work.
00:59 < needs1> Still wondering if "/players?p=1" as a shorturl for "/players/by-rank?p=1" is a good choice...
01:05 <@minus> Store historical data, that's always good to have later 
19:35 <@minus> heinrich5991: did you implement that http master stuff?
19:51 <@heinrich5991> mh. haven't you seen the link I posted a couple of days ago?
19:52 <@heinrich5991> https://github.com/teeworlds/teeworlds/compare/master...heinrich5991:http_master7
19:52 <@minus> ah, was looking through pull requests
19:52 <@heinrich5991> minus: ^
19:53 <@minus> https://github.com/teeworlds/teeworlds/pull/1395 obsolete?
19:53 <@heinrich5991> if that is fine by Oy, I'd port it to 0.6
19:53 <@heinrich5991> eh, that's an old version of the same stuff
19:53 <@minus> close it then
19:54 <@minus> i'd first integrate it into 0.7, wouldn't you?
19:54 <@heinrich5991> I would integrate it into 0.6 because we have the problems there, right now
19:55 <@minus> fair enough
19:55 < EastByte> how do you keep support for old clients?
19:55 < EastByte> and servers
19:55 <@heinrich5991> keep the old master servers running, as slaves of the new ones
19:56 <@heinrich5991> if they go down, old clients and servers have problems, but not new ones
19:56 < EastByte> okay
19:56 <@heinrich5991> should provide a little incentive to upgrade as well
19:57 <@minus> backwards compat sucks
19:57 < EastByte> well and once it's merged into ddnet it should spread pretty fast
19:58 <@heinrich5991> we do have a lot of servers that won't be upgradeable easily
19:58 <@heinrich5991> I haven't figured out a solution for these yet
19:58 <@minus> 0.7 is one
19:59 <@heinrich5991> that won't help either :)
19:59 < EastByte> so the plan is to use a CDN, there is still the fact that it might not work with cloudflare because of browser validation
19:59 <@heinrich5991> that's something you can disable IIRC
20:00 <@heinrich5991> or rather, have to explicitly enable
20:00 <@minus> lemme check on that
20:00 < EastByte> not sure about that in the free version
20:00 < EastByte> disabling it will increase cost on cf's side
20:01 <@minus> oh great, my account broke cloudflare again
20:01 < EastByte> well, but there are a few alternatives anyways
20:02 <@minus> EastByte is right, turning the captchas off requires an enterprise plan
20:03 <@heinrich5991> meh, I had a site under cloudflare once, need to get one again
20:03 <@minus> >$5,000+/month
20:03 < EastByte> 'enterprise' :p
20:03 <@minus> >Essentially off: Challenges only the most grievous offenders
20:04 < EastByte> ovh cdn info page seems to have problems with itself: https://www.ovh.de/cdn/
20:04 <@heinrich5991> that sounds like off, no?
20:08 <@minus> hah
20:16 <@minus> seems it's possible to disable the browser verification stuff
20:16 < EastByte> cf?
20:17 <@minus> yeah
20:17 < EastByte> well, try it out
20:18 <@heinrich5991> they don't enable it by default at least
20:18 <@heinrich5991> I was able to test their caching
20:19 <@minus> speaking of caching
20:19 <@minus> we should include a timestamp in the server list
20:19 <@heinrich5991> why?
20:20 <@minus> so we can tell how old the list is
20:20 <@minus> helpful for debugging and if we use CF's always online
20:21 <@matricks> can someone recommend a good movie? :P
20:21 <@minus> the man from earth
20:22 <@matricks> the one from 2007?
20:22 <@minus> i think, yeah
20:23 <@matricks> hmm, hot singles in my area
20:23 <@minus> a word of warning: the whole movie takes place in a single room pretty much
20:23 <@matricks> minus: no worries, Ive seen cube
20:23 <@minus> haha
20:24 <@matricks> good movie
20:24 <@minus> another movie i've recently watched is limitless (2011), that was also quite enjoyable
20:24 <@minus> dunno about cube
20:24 <@matricks> think I've seen that one
20:24 <@minus> been a loooong time since i've maybe seen that
20:25 <@matricks> I can recommend Predestination
20:25 <@minus> limitless is about a drug that makes you super intelligent pretty much
20:25 <@matricks> yeah... might have seen it
20:25 < Stitch626> too
20:26 <@minus> predestination sounds interesting
20:26 <@matricks> it's freaking awesome
20:26 <@matricks> saw that movie over a year ago and I can't let it go
20:27 <@minus> oh
20:27 <@minus> i already have predestination
20:27 <@minus> no wonder the imdb link was marked as visited
20:28 <@minus> >// NOTE: This looks very wrong.
20:28 <@minus> nice comment, heinrich5991
20:29 <@minus> dualstack handling in teeworlds is a mess
20:30 <@minus> i dare say
20:30 <@matricks> yap
20:31 <@matricks> I... I kinda wanna watch predestination again...
20:31 < Stitch626> i think i will too next 1-2 days
20:31 <@matricks> Stitch626: seen it?
20:31 < Stitch626> mb, i always forget about movies i've seen
20:31 <@minus> seeing how much state'y code teeworlds has, has anyone ever thought about using some thing that handles state machines and gives you a nicer interface to them? like ragel https://www.colm.net/open-source/ragel/
20:32 < Stitch626> so i can see the same movie up to 5 times
20:33 <@minus> i've been working on a C++ lib (i.e. looking at a possible design) with predefined states, statechanges and callbacks (mainly intended for lambdas)
20:34 <@matricks> probably is something for it in boost ;)
20:34 <@minus> haha
20:35 <@minus> 2 things actually
20:36 <@matricks> now, movie
20:37 <@minus> predestination?
20:37 <@matricks> limitless
20:37 <@minus> that was fast
20:37 <@matricks> took 3 minutes to get or someting
20:37 <@minus> internet too slow to stream :(
20:37 <@matricks> got 250mbit connection
20:38 <@minus> damn you swedes
20:38 <@matricks> :D
20:38 <@minus> been stuck on the same shitty DSL connection for over 10 years
20:39 <@minus> http://www.boost.org/doc/libs/1_59_0/libs/statechart/doc/tutorial.html compile-time everything, interesting
20:40 <@minus> my approach is similar but run-time everything
20:40 <@minus> (state transitions still predefined)
20:50 <@minus> actually Boost's Meta State Machine library is even more compilery; huge template-based tables
22:15 <@matricks> minus: drugs are good m'key
22:15 <@minus> uhm, okay
22:15 <@matricks> limitless
22:16 <@minus> right
22:16 <@minus> watching predestination right now
22:17 <@minus> how'd you like the ending, matricks?
22:17 <@matricks> meh
23:09 <@minus> finished predestination, nice
23:10 <@matricks> I think it's very cool
23:10 <@matricks> really nice layout of the movie
23:10 <@minus> at ⅔ through the title clicked and i knew how it would turn out
23:15 <@matricks> yeah, but I think it's really beautifully done
23:15 <@matricks> how everything ties together
23:25 <@minus> should've tied in robertson for maximum effectiveness