Time  Nick        Message
07:19 kahrl       hmmmm: actually httpfetch is 'kind of' ready
07:19 kahrl       see http://irc.minetest.ru/minetest-dev/2013-11-04#i_3412316
07:21 kahrl       a problem is that the serverlist part of it has not been tested much since the plan is to use async_events for that purpose
07:21 VanessaE    kahrl: point of interest:  without http_fetch, I can pull data from my VPS at ~135 kB/sec.
07:22 VanessaE    I have not tested it with, yet.
07:22 kahrl       so if we want to fix the serverlist stall we'd have to merge async_events, not httpfetch (or test the serverlist part of httpfetch, merge that for 0.4.8, and replace it with the async_events variant for 0.4.9)
07:22 kahrl       VanessaE: that is with curl though, right?
07:22 VanessaE    yes
07:26 VanessaE    without, Pilzadam says he was barely able to sustain 5 kB/sec :)
07:28 VanessaE    foe the 0.4.9 milestone, I'd like to suggest caching all models and other media, client side, that isn't so now.
07:28 VanessaE    for*
07:28 kahrl       hey, that's better than a 9600 baud modem
07:28 VanessaE    heh
07:29 kahrl       all media is cached
07:29 kahrl       the downloading code doesn't make a distinction between file types
07:29 VanessaE    to me, "media" includes models also
07:29 kahrl       well those are cached too
07:29 VanessaE    no
07:29 VanessaE    they aren't
07:30 kahrl       what you're seeing is a loading bug
07:30 VanessaE    someone of authority, also PA I think, has said they aren't being cached?
07:30 VanessaE    interesting
07:30 kahrl       I don't remember the details of the bug though
07:30 VanessaE    no matter
07:31 VanessaE    when either httpfetch or async goes in, that point will be moot :)
07:31 kahrl       (there's no if (!string_ends_with(filename, ".x")) add_to_cache() in the downloading code)
07:32 kahrl       well httpfetch yes, async_events doesn't have anything to do with it
07:32 VanessaE    there's been the claim that async will fix that loading glich...somehow
07:32 VanessaE    I don't remember the details either
07:32 kahrl       huh
07:33 kahrl       the bug happens when the client is already running... at that point the mainmenu lua stack has already been destroyed, async_events shouldn't be around anymore
07:34 kahrl       unless I'm misunderstanding the design
07:34 VanessaE    beats me - you already know more about it than I do :P
07:35 VanessaE    ok, time to give my brain a rest.
07:35 VanessaE    night.
07:35 kahrl       night
14:23 BlockMen    i know we have a freeze, but does that affect fixes like this?  https://github.com/BlockMen/minetest/commit/0e6a184809c5d1c38963f5862930b8089f06fb7b
14:35 ShadowNinja BlockMen: Simple fixes are still ok. That looks good to me.
14:36 ShadowNinja (But can paths actually contain characters that need to be escaped?)
14:36 BlockMen    ShadowNinja, "/". actually it loops through and says "cant open ....
14:39 BlockMen    ShadowNinja, if you agree i gonna push it then
14:40 ShadowNinja BlockMen: That shouldn't cause any issues, maybe '\' though. If it works then I'm OK with it.
16:38 PilzAdam    I got another build failed e-mail from launchpad: https://launchpad.net/~minetestdevs/+archive/daily-builds/+recipebuild/586611
16:42 ShadowNinja I don't see how that build log is even remotely helpfull. Maybe Trusty has a bug?
17:21 PilzAdam    ShadowNinja, I havent looked at the log yet, Im just passing it into this channel
17:24 sapier      vanessa async doesn't do anything after mainmenu is destroyed for the moment
17:25 VanessaE    sapier: ok, well someone (I forget who) seemed to be convinced that adding the async stuff would fix the load-order issue with some files coming off remote_media
17:25 sapier      there are plans to add async support to scriptapi too but I need to fix the environment locking issues first, async wont be very helpfull without it
17:26 sapier      no async is what it's name tells async things only :-) that's why there's no conflict to httpfetch too
17:27 VanessaE    then httpfetch IS needed to fix that load order bug
17:28 sapier      and I don't like the somehow async httpfetch addons as they are specific for downloading and can't be used for anything else ... while async is designed to be used for everything (once some basic core issues are fixed)
17:28 sapier      while I am absolutely positive about the httpfetch core
17:29 sapier      https://github.com/sapier/minetest/tree/async_events_plus_httpfetch is async using httpfetch
17:30 sapier      but I'm not 100% sure kahrl didn't fix things after I merged httpfetch
17:31 sapier      see httpfetch as backend while async is frontend to scripting
17:36 VanessaE    well either way there needs to be less hemming-and-hawing and more tuning-and-committing.  If the part of the code that is needed to fix this bug works properly and works well, wtf is the delay?
17:37 sapier      what bug are we talking about?
17:38 VanessaE    Certain people people kept hok me a tchynik about using remote_media.  ow that I use it on my servers, it revealed this out-of-order resource loading issue that causes clients to sometimes miss image files or models...and now no one wants to JUST FIX IT. :P
17:38 VanessaE    now*
17:39 sapier      ohh that one :-)
17:41 sapier      OldCoder I found your nil deserialization, you can ignore it it's a initializer trying to initialize with nil fixing a invalid initialization after that
17:42 VanessaE    (insert random "Yo Dawg..." meme here)
17:47 sapier      :-) I know this doesn't help but I guess that bug doesn't happen enough often
17:47 sapier      btw when is release planed to be?
17:47 VanessaE    it happens on probably any server out there that uses remote media.
17:48 VanessaE    no idea
17:49 PilzAdam    sapier, next weekend?
17:49 sapier      ok wasn't sure remembered something about tuesday
17:50 celeron55   launchpad... blah
17:51 celeron55   debian has changed something in something related to packaging and then there's that obscure debian-specific issue "native package version may not have a revision" in it
17:51 celeron55   i have no way of knowing what "native package version" means and neither do i have any idea why it shouldn't have a revision and i don't have any clue how it's supposed to be done differently
17:52 celeron55   i could remove {revno} from it, but i don't know what the consequences from it are
17:53 celeron55   this is why i don't want to do anything distro-specific in minetest...
17:53 celeron55   they think everybody in the world know their internal things
17:53 celeron55   it's plain stupid
17:54 OldCoder    sapier, so any fix would be cosmetic?
17:54 sapier      yes
17:55 OldCoder    sapier, Thank you. My mindset is such that I prefer to fix things; I may look for a cosmetic fix.
17:55 OldCoder    sapier, thanks for (1) this information (2) the cows setting change (3) the information about the load time lockup
17:55 OldCoder    You have been most helpful
17:56 celeron55   apparently "native package" means something like a package that is used only in debian
17:56 * OldCoder  will test related to load time lockup and looks forward to a fix for that
17:56 celeron55   but this for sure isn't that
18:05 celeron55   http://i.imgur.com/xbNvRPY.png
18:05 celeron55   yeah no thanks, trying to understand this is an absolute waste of time
18:06 celeron55   someone go drag into here some creature that understands something about debian packaging and launchpad
18:07 celeron55   those probably live in deep rocky and muddy holes
18:10 OldCoder    celeron55, I am pleased to see that you are in a better mood than usual. BTW I cannot core dev ATM but hope to do so as things proceed.
18:11 OldCoder    celeron55, I am sure that the "creatures" from muddy holes will be pleased to come. I will ask around.
18:20 sapier      "but static data doesn't actually exist in (4,-2,-41)" shouldn't those things be fixed with the object duplication fixes?
20:20 artur99_    !seen oldcoder
20:32 sfan5       ^ lol